path: root/ChangeLog

2021-11-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Make allocator replacement optional.  Switch it
	off by default, relying on C++ runtime new/delete support.

	* configure.ac: Call new `GROFF_USE_GROFF_ALLOCATOR` m4 macro.
	Use `AM_CONDITIONAL` to set Automake variable
	`USE_GROFF_ALLOCATOR` if appropriate.  Report whether the
	allocator is used in configure script output summary.
	* m4/groff.m4 (GROFF_USE_GROFF_ALLOCATOR): Define new macro to
	collect user preference.  The default is off.
	* src/libs/libgroff/libgroff.am (libgroff_a_SOURCES): Build and
	link new.cpp only if `USE_GROFF_ALLOCATOR`.

	* NEWS: Add item.

2021-11-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	* m4/groff.m4 (GROFF_TMAC): Report a human-readable message if
	no system tmac prefix is found, instead of leaving the ellipsis
	hanging.

2021-11-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	* m4/groff.m4 (GROFF_TMAC): Use separate shell variable for
	`AC_MSG_RESULT()` content since the human-readable string we
	populate it with in the empty case is not appropriate for later
	`AC_SUBST()`-itution.  Problem introduced by me in commit
	19670348, 31 October.

2021-11-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tmac.am (MOSTLYCLEANFILES): Remove `$(TMACMDOCFILES)`.
	This variable, formerly named `TMACMDOCSTRIPFILES`, should not
	be used in a clean target because its contents are (no longer)
	generated files.  Problem introduced by me in commit 24602f42, 4
	July (only noticeable in within-source-tree builds).

2021-11-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff,pic]: Check `strdup()` return value.

	* src/libs/libgroff/font.cpp (struct text_file): Add `fatal()`
	member function.
	(text_file::fatal): Implement it.

	* src/libs/libgroff/font.cpp (font::load_desc):
	* src/preproc/pic/troff.cpp (troff_output::set_location): Die if
	`strdup()` returned a null pointer.

2021-11-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libgroff/error.cpp (do_error_with_file_and_line):
	Revise to eliminate `fprintf()` calls, which might perform
	dynamic memory allocation, rendering this function unsafe to
	call after memory allocation failures.  Since this function is
	near the top of our diagnostic output call stack, that would be
	unfortunate.  `errprint()` does not use dynamic allocation, nor
	do the `i_to_a()` and `ui_to_a()` functions it calls to format
	integers.

2021-11-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	* configure.ac: Add `strdup` to AC_CHECK_FUNCS since we are
	using it and the whole point of libgroff's `strsave()` was to
	get along without it.  But `strdup` has been standardized in
	POSIX for 20 years (SUSv3, POSIX:2001) and is on its way into
	ISO standard C (N2354) and C++ (P2391R0) as well.

2021-11-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libgroff/strsave.cpp (strsave): Call `strcpy()` only
	if `malloc()` did not return a null pointer.  Problem noted by
	Ingo Schwarze.

2021-11-07  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libgroff/fontfile.cpp (font::open_file): Don't open
	user-specified font file names with slashes in them; i.e., don't
	traverse directories outside the configured font path.  Also
	refuse to open the file if the `sprintf()` used to construct its
	file name doesn't write the expected quantity of bytes to the
	destination buffer.

	Fixes <https://savannah.gnu.org/bugs/?61424>.  Thanks to Ingo
	Schwarze for feedback.

2021-11-07  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Regression-test Savannah #61424.

	* src/roff/groff/tests/fp_should_not_traverse_directories.sh: Do
	it.
	* src/roff/groff/tests/artifacts/HONEYPOT: Add test artifact.
	* src/roff/groff/tests/artifacts/devascii/README: ...and this;
	we need an empty directory to make the test work but such things
	tend to look unintentional.
	* src/roff/groff/groff.am (groff_TESTS): Run test.
	(EXTRA_DIST): Ship artifacts.

2021-11-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Fix diagnostic error wording.

	* src/libs/libgroff/font.cpp (font::load): Fix diagnostic
	message; this is issued when an unrecognized directive is
	encountered, so it necessarily occurs _before_ any `charset` or
	`kernpairs` directive.  Give the user a hint accordingly.

2021-11-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Fix paragraph tag regressions.

	* tmac/an-ext.mac (UR, MR): Only create an environment and
	diversion if we're not already in one of the latter.  If we are
	{in groff man, this occurs only when collecting a `TP` paragraph
	tag}, then typeset the URI as part of the indented paragraph.
	(UE, ME): Pop the diversion and environment stacks only if we
	pushed them in the first place.  Eliminate spurious space in
	post-URI arguments by only typesetting excess macro parameters
	if there were any.

2021-11-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Regression-test Savannah #61425.

	* tmac/tests/an-ext_link-macros-work-in-paragraph-tags.sh: Do
	it.
	* tmac/tmac.am (tmac_TESTS): Run test.

2021-11-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac (DS): Drop a redundant unconditional break; every
	display macro we call in the event of a valid call already
	breaks the line, and it's polite to not break if we're given an
	invalid argument.  Emit a diagnostic error if the argument we're
	given is invalid, and immediately end the diversion we just
	opened.

2021-11-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/env.cpp (environment::set_font): Clarify
	diagnostic warning.

2021-11-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/table.cpp (table::add_entry): If we see a '\^'
	entry on the first row of a table, it's invalid, but we need to
	create an empty entry in its place.  Otherwise, someone can put
	another '\^' right below the one on the first row, creating a
	reference to a nonexistent table entry and provoking a SEGV.
	Issue an error diagnostic (distinct from the one in
	`do_vspan()`), create the entry, and skip `do_vspan()` (given a
	'^' in a first-row definition, it issues an error diagnostic and
	returns early, which suffices).  Problem appears to date back to
	groff 1.02 (June 1991) at the latest.

	Fixes <https://savannah.gnu.org/bugs/?61417>.

2021-11-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	[tbl]: Regression-test Savannah #61417.

	* src/preproc/tbl/tests/\
	do-not-segv-on-invalid-vertical-span-entry.sh: Do it.
	* src/preproc/tbl/tbl.am (tbl_TESTS): Run test.

2021-11-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Fix Savannah #61408.

	* tmac/an.tmac (an-prepare-inner-footer): Don't try to get a
	substring of an empty string; the man page author might not have
	specified a third argument to `TH` (or supplied an empty one).

	Fixes <https://savannah.gnu.org/bugs/?61408>.  Thanks to Bjarni
	Ingi Gislason for the report.

2021-11-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Add regression test for Savannah #61408.

	* tmac/tests/an_inner-footer-abbreviation-works.sh: Do it.

2021-11-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Tweak computation in URI breaking.

	* tmac/an-ext.tmac (UE, ME): Stop adding the page offset `.o`
	when performing the available horizontal space computation for
	the typeset URI string.  This made no difference on nroff
	devices (grotty; even with `.po 15n`), but it caused the
	breaking decision to be too conservative on troff devices
	{grops}, forcing some URIs that would fit on the current line to
	the next one.

	* tmac/tests/an-ext_UE-breaks-before-long-URIs.sh: Add test of
	URI with no link text, which also has break points after every
	character, so that we detect even slight alterations.

2021-11-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Do more device and font description file validation,
	resolve an assertion failure arising from a negative declared
	device resolution, and correct a documentation error.

	* src/libs/libgroff/font.cpp (font::load): Include more
	information about invalid input in diagnostic messages.
	  - When a kern pair's amount is missing or invalid, report the
	    name of the kern pair.
	  - When someone tries to declare the first entry in the charset
	    section as an alias, report the glyph name.
	  - Identify the token for the unnamed character if an attempt
	    is made to alias it.
	  - When an out-of-range character type is applied to a glyph,
	    name the glyph.
	(font::load_desc): Same.
	  - Drop redundant zero initialization of `res`.
	  - Check all directives that take basic units for positive
	    values, adding `res`, `unitwidth`, `paperwidth`, and
	    `paperlength`.  Update this diagnostic to demand positive,
	    not nonnegative, values.
	  - When the font count is long in a `fonts` directive, report
	    how many font names were declared (and thus expected).  (If
	    the count is short, the next line is read for a font name,
	    like 'tcommand' in our devutf8/DESC.)
	  - When interpreting a `papersize` directive, throw an error
	    and return false if `res` has not yet been encountered,
	    since it is used in subsequent computations.
	  - When a paper format cannot be determined, report the
	    original declared value from the DESC file.  Use `strdup()`
	    to save it since it gets clobbered by the resolving process.
	    `free()` the saved string when we're done, regardless of
	    error condition.

	* doc/groff.texi (Device and Font Files):
	* man/groff_font.5.man (DESC file format): Document additional
	exception to order-indifference of directives: (at least one)
	`res` must precede `papersize`.

	Fixes <https://savannah.gnu.org/bugs/?61414>.

2021-11-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/devices/grohtml/post-html.cpp
	(html_printer::do_file_components): Add assertion.

2021-11-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/devices/grops/ps.cpp (usage): Document -p and --version
	options.

2021-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/mdoc/doc-nroff (doc-setup-page-layout): Set the title
	length to the device line length (register `.l`) if register
	`LT` is not defined.  This keeps the title length from being set
	to 78n when the line length is not, preventing the output from
	looking silly.

2021-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/main.cpp (main): Emit groff code to define
	macro `T&` as empty if it is not already defined.  (All of our
	full-service macro packages [except mdoc(7)] already do this, so
	there's no change in semantics; mom(7) even defines it as an
	empty _string_.)

2021-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/tests/string_case_xform_requests.sh: Migrate
	test to POSIX shell, dropping use of process substitution (a
	Bashism).  Also rewrite to stop using a here document within a
	command substitution, mainly for paranoia.

	* src/roff/groff/tests/string_case_xform_requests.sh:
	* src/roff/groff/tests/on_latin1_device_oq_is_0x27.sh: Update
	shebang lines to use /bin/sh as the interpreter.

2021-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/devices/grops/ps.cpp (encode_subfont): Update assertion to
	check for `sub` parameter being a null pointer rather than
	`sub->glyphs`; the latter is a member array of a `subencoding`
	struct so it can't be a null pointer.  (`glyphs` is an array of
	pointers to `const char`; the _elements_ of the array can [and
	are initialized to] be null pointers, but the address of the
	array itself will never be in a C/C++ implementation.)  Detected
	by Clang 13's "-Wtautological-pointer-compare".

2021-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libgroff/new.cpp (delete): Declare `throw()` (no
	exceptions thrown); quiets "-Wimplicit-exception-spec-mismatch"
	from Clang 13.

2021-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* m4/groff.m4: Fix messaging nits.
	(GROFF_MAKEINFO): Set `missing` to "missing 'makeinfo'" for
	consistency with other assignments to `missing`.
	(GROFF_TMAC): Utter an intelligible result in the configure
	check when no system macro files requiring groff wrapping are
	located.

2021-10-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Handle special character escape sequences that map to
	basic Latin glyphs in device control escape sequences
	consistently among output devices.

	* src/roff/troff/input.cpp (encode_char): Rearrange
	conditionals.  This is the logic that puts the "whatever" within
	a \X'whatever' escape sequence into GNU troff's intermediate
	output.  Handle stretchable and unstretchable space escape
	sequences ("\ " and \~") first.  Then, if the token is a special
	character escape sequence, retrieve its "contents" (glyph name).
	Move the basic Latin mapping for the seven glyph names '-',
	'aq', 'dq', 'ga', 'ha', 'rs', and 'ti' here, before checking
	whether the device description issued the
	'use_charnames_in_special' directive.  This way, the 'html' and
	'xhtml' output devices can straightforwardly embed these basic
	Latin characters in device control escapes (notably, "html:",
	for which the present convention is to follow the this tag
	immediately with a literal HTML URI, complete with `<a href>`
	element syntax).  If the special character is none of these and
	we should 'use_charnames_in_special', proceed as groff 1.22.4
	and earlier did.  This is a behavior change, as was my addition
	of this translation mechanism in the first place, so...

	* doc/groff.texi (Postprocessor Access): Document it.

	* src/roff/groff/tests/\
	device_control_escapes_express_basic_latin.sh: Test it.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

	Fixes <https://savannah.gnu.org/bugs/?61401>.

2021-10-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Map \[ti] correctly in device control escape sequences.

	* src/roff/troff/input.cpp (encode_char): Fix copy-and-paste
	error.  \[ti] should put '~', not '^', into a device control
	command.

	Fixes <https://savannah.gnu.org/bugs/?61403>; problem introduced
	by me in commit 9d61b3d1, 1 October.

2021-10-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Handle degenerate input quietly.

	* tmac/an.tmac (TH): Define new register `an-TH-was-called`.
	(an-end): Return immediately if that register is not defined;
	to format the default page footer we must have the information
	declared in a valid `TH` call.  (`TH` also initializes the type
	size and baseline spacing registers we use to prepare the page
	footer environment.)  If the register _is_ defined, remove it
	just prior to the end of this macro definition, in preparation
	for next page to be rendered.

	* tmac/tests/an_handle-degenerate-input-quietly.sh: Test it.

	Fixes <https://savannah.gnu.org/bugs/61402>, a regression from
	groff 1.22.4 (problem introduced by me in the course of many
	changes to trap management and header/footer handling to work
	nicely in batch rendering with -mandoc and mdoc(7) documents).

2021-10-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Warn if `TE` table macro called but `TW` register (set by
	tbl(1)) undefined.  Arrange it so that we warn only once per
	man(7) document, not per table.

	* tmac/an.tmac (TH): Remove `TW` register when processing new
	document and clear `an-was-tbl-failure-reported` register.
	(TE): Check `TW` and `an-was-tbl-failure-reported` registers; if
	the former is undefined and the latter false, emit the message
	and set `an-was-tbl-failure-reported`.

	Addresses part of <https://savannah.gnu.org/bugs/?61379>.
	Thanks to Bjarni Ingi Gislason for the suggestion.

2021-10-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	[tests]: Improve portability.

	* src/devices/grotty/tests/osc8_works.sh: Fix typo in test.
	Multiple regex quantifiers after an atom were not flagged with a
	diagnostic by GNU grep 3.3, but were by BSD grep 2.5.1-FreeBSD.

	* src/roff/groff/tests/smoke-test_html_device.sh: Stop trying to
	set $LC_CTYPE to "C.UTF-8"; some systems don't support this
	expediency.  Skip the test if the tester hasn't configured the
	environment adequately.

	* tmac/tests/s_TC-works-with-percent-in-custom-titles.sh: Fix
	portability problems exposed by FreeBSD sed (which version is
	unclear).  {1} Use POSIX BREs, dropping use of '+' quantifier.
	{2} Add semicolons to terminate commands before braces on the
	same line.  Simplify surrounding test structure.
	* tmac/tests/e_footnotes-work-with-columns.sh: Fix same two sed
	problems.
	* tmac/tests/andoc_flush-between-packages.sh:
	* tmac/tests/doc_accept-mixed-case-section-headings.sh:
	* tmac/tests/an_TS-do-not-keep-tables-when-cR-set.sh: Fix same
	sed semicolon problem.

	* src/roff/groff/tests/on_latin1_device_oq_is_0x27.sh: Rewrite
	test to stop using a here document (containing an unpaired
	single quote) inside a command substitution, which is broken in
	GNU Bash 3.2 and some other versions.  Instead construct the
	input with printf(1).  Also gets rid of a Bashism (process
	substitution) that Ingo pointed out years ago.
	* src/utils/grog/tests/recognize-perl-pod.sh: Same problem,
	different solution.  The Perl POD output is far too large to
	construct programmatically, so move it into an external file...
	* src/utils/grog/tests/foo.man: ...here.
	* src/utils/grog/grog.am (EXTRA_DIST): Ship new test artifact.

2021-10-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/doc.am: Make sed-based insertion of `lf` requests more
	portable: a dance is required to embed a literal newline inside
	a sed expression inside a shell command executed as part of a
	Make target rule.  We assume that neither the POSIX shell $''
	operator nor a working printf(1) is available.
	(DOC_GROFF): Define a shell variable containing a literal
	newline, evading make(1) and echo(1)'s valiant attempts to
	consume it.
	(doc/pic.html, doc/webpage.html): Repeat trick, slightly
	modified since the dependency needs to be manually identified.

	Solution adapted from the GNU Autoconf manual, "Newlines in Make
	Rules".

	Fixes <https://savannah.gnu.org/bugs/?61265>.  Thanks to John
	Gardner for the report.

2021-10-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an.tmac (TH): Stop populating `an-extra3` (the default
	center header) if the section argument is "3p".  "3p" does not
	always mean a Perl-related man page; Debian- and Arch
	Linux-based systems[1][2] (at least) have long used a "3perl"
	suffix, while OpenBSD appears to still use "3p" for Perl[3],
	which is also used by the POSIX man page distribution[4].
	Fortunately, Perl and POSIX are reliable at providing a fifth
	argument to `TH` anyway, so our fallback is as unnecessary as it
	is non-impartial.
	* tmac/groff_man.7.man.in (Document structure macros) <TH>:
	De-document the above behavior.

	[1] https://manpages.debian.org/bullseye/perl-doc/\
	Pod::Man.3perl.en.html
	[2] https://man.archlinux.org/man/Pod::Man.3perl
	[3] https://man.openbsd.org/man3p/Pod::Man.3p
	[4] https://man7.org/linux/man-pages/man3/system.3p.html

2021-10-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Abbreviate the inside footer if necessary.

	* tmac/an.tmac (an-prepare-inner-footer): Prevent the inside
	footer from overrunning the center footer.
	(BT): Use new string `an-ifoot` instead of `an-extra2` in page
	footers.

	Fixes <https://savannah.gnu.org/bugs/?61386>.

2021-10-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #61386.

	* tmac/tests/an_inner-footer-abbreviation-works.sh: Do it.
	* tmac/tmac.am (tmac_TESTS): Run test.

2021-10-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff_char.7.man (Description, Special character escape
	forms): Document the `\C` escape sequence.  Fix error: \[a] is
	_not_ "translated to \a, the uninterpreted leader escape
	sequence," but requests a glyph _named_ '\a' internally--it
	appears thus in diagnostic messages and, importantly, in the
	font description files defining special character glyph names.

2021-10-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an.tmac: Refactor to reduce repetition.
	(an-break-paragraph): New macro handles paragraph-breaking tasks
	previously duplicated.
	(SH, SS, P, TP, IP, HP): Call it rather than issuing its
	constituent requests.

2021-10-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an.tmac (TP): Reset type size, vertical spacing, and font
	to defaults before setting the new paragraph.

	Fixes <https://savannah.gnu.org/bugs/?61380>.

2021-10-24  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Auto-increment): Fix error; the `\R` escape
	sequence does _not_ support an auto-incrementation amount.
	* man/groff_out.5.man (Command reference/Simple commands): Fix
	erroneous claim: 'N' is not a GNU roff extension; it appears on
	page 27 of CSTR#54 (1992).

2021-10-24  Keith Marshall  <keith.d.marshall@ntlworld.com>

	[ms]: Provide global default XH and XN implementations.

	cf. <https://savannah.gnu.org/bugs/?58946#comment13>

	* tmac/s.tmac (XH-INIT, XN-INIT, XH-UPDATE-TOC)
	(XH, XN, XH-REPLACEMENT, XN-REPLACEMENT): Implement them, and...
	* tmac/groff_ms.7.man: ...document them.

2021-10-24  Keith Marshall  <keith.d.marshall@ntlworld.com>

	[ms]: Fix misleading NH macro indentation.

	* tmac/s.tmac (@NH) [.T is html]: .if statement should exhibit
	same indentation as preceding .DEVTAG-NH statement, but is
	overindented; fix it.

2021-10-23  Keith Marshall  <keith.d.marshall@ntlworld.com>

	[ms]: Defend against uncontrolled page trap recursion.

	* tmac/s.tmac (pg@top) [HM+FM+.V>.p]: Diagnose insufficient page
	length, and abort; cf. <https://savannah.gnu.org/bugs/?61294>

2021-10-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	[groff]: Fix code style nits.

	* src/roff/groff/groff.cpp (main, help): Use standard C library
	preprocessor symbols `EXIT_SUCCESS` and `EXIT_FAILURE` instead
	of 0 and 1 literals, respectively.  See commit fa4c27e9, 7
	September.
	(main): Replace `assert(0)` with a communicative predicate.

2021-10-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/devices/grotty/tty.cpp (tty_printer::special_link):
	Eliminate casts by retyping `uri` and `pair` to pointers to
	`const char`s.

2021-10-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tmac.am: Build more quietly.
	(tmac/stamp-wrap): Prefix all rule commands with `$(AM_V_at)`;
	since this target is a stamp file instead of a useful artifact
	to be distributed, make its creation completely silent by
	default.

2021-10-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/troff.am: Build more robustly.
	(src/roff/troff/majorminor.cpp): Drop progress-reporting `echo`
	in favor of prefixing final rule command with `$(AM_V_GEN)`.
	Prefix earlier rule commands with `$(AM_V_at)`.  Use `printf`
	instead of `echo` to build the file, and create a temporary file
	at first so that we don't race with the compiler on parallelized
	builds.

2021-10-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/eqn/eqn.am (neqn): Simplify: drop removal of
	target prior to overwriting it; any system we can build groff on
	should have a shell that implements the clobbering semantics of
	the POSIX shell '>' operator.  (If this isn't true, we have
	_many_ other target rules in our build to which we must add
	prefatory `$(RM)`s, or I miss my guess.)

2021-10-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libgroff/libgroff.am: Build more quietly and
	robustly.
	(src/libs/libgroff/version.cpp): Drop progress-reporting `echo`
	in favor of prefixing final rule command with `$(AM_V_GEN)`.
	Prefix earlier rule commands with `$(AM_V_at)`.  Use `printf`
	instead of `echo` to build the file, and create a temporary file
	at first so that we don't race with the compiler on parallelized
	builds.  Drop nilpotent sed transformation; this looks like a
	relic of past times when we'd drop a trailing ".0" from the
	version number stored in the C symbol `Version_string`.
	(charset.alias, ref-del.sed, ref-add.sed): Prefix rule command
	with `$(AM_V_GEN)`.

2021-10-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/devices/xditview/xditview.am: Build more quietly.
	(src/devices/xditview/GXditview-ad.h): Drop progress-reporting
	`echo` in favor of prefixing rule command with `$(AM_V_GEN)`.

2021-10-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* font/devpdf/devpdf.am: Build more quietly.
	(font/devpdf/enc/text.enc, font/devpdf/map/symbolmap,
	$(DEVPDFFONTMAP_1)): Prefix all rule commands with
	`$(AM_V_at)`; since these are all `mkdir`s and `cp`s, I reason
	that little can go wrong with them that won't be disclosed by
	these tools' own diagnostic messages, so they are insignificant
	from a perspective of build progress.

2021-10-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/doc.am: Build more quietly.
	(doc/examples.stamp): Prefix all rule commands with
	`$(AM_V_at)`; since this target is a stamp file instead of
	a useful artifact to be distributed, make its creation
	completely silent by default.
	(doc/groff.info, .texi.txt): Prefix directory creation with
	`$(AM_V_at)` and `makeinfo` command with `$(AM_V_GEN)`.
	(.texi.dvi, .texi.pdf): Prefix directory creation with
	`$(AM_V_at)` and `makeinfo` command with `$(AM_V_GEN)`.  Send
	diagnostic message to standard error if `texi2dvi` is
	unavailable.
	(.texi.html): Prefix directory creation with
	`$(AM_V_at)`.  Two HTML versions are produced, in split and
	unsplit versions; prefix one `makeinfo` command with
	`$(AM_V_GEN)` and the other with `$(AM_V_at)` so that only one
	message is produced.
	(gnu.eps): Drop useless/noisy `echo` command.  Prefix copy
	operation with `$(AM_V_at)`.  Send diagnostic messages to
	standard error if there are problems with the Netpbm tools.

	(MOSTLYCLEANFILES): The `mostlyclean` target was leaving a file,
	`groff.info-3`, behind in the build tree.  At some point our
	Texinfo manual grew beyond earlier expectations.  Use a glob
	`[0-9]` instead (giving a hostage to fortune that `makeinfo`
	will never split our manual into more than nine chunks).

2021-10-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* arch/misc/misc.am:
	(arch_shdeps_sh): Pull file name into a new variable.
	(EXTRA_DIST, shdeps.sed): Ue it.
	(shdeps.sed): Build more quietly; prefix with `$(AM_GEN_V)`.

2021-10-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	[nroff]: Support groff's -R flag to run refer(1).

	There's no reason not to support calling `refer` with nroff-mode
	documents; typographically, it's less demanding than tbl(1).
	GNU troff(1) supports an incompatible `-R` flag to suppress the
	reading of troffrc{,-end} files, but this is likely not the `-R`
	that users will want--it is most useful for troubleshooting
	troff, and nroff already unconditionally loads a macro file
	{tty-char.tmac}.

	* src/roff/nroff/nroff.sh: Do it.

	* src/roff/nroff/nroff.1.man:
	* NEWS: Document it.

2021-10-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tmac.am (tmac/groff_man.7.man,
	tmac/groff_man_style.7.man): Be quieter by default; use
	$(AM_V_GEN) more consistently with the rest of the build.

2021-10-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (do_error): Format diagnostic
	messages in closer alignment with GNU Coding Standards: don't
	introduce a space between the program name and the input file
	name.

	* src/roff/groff/tests/string_case_xform_errors.sh: Update
	expected output.

2021-10-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/env.cpp (distribute_space): Revert an
	`assert()` I added in commit b93eacd8d7 (5 September); we can
	indeed reach this code with a negative amount of desired space,
	and in fact the "show hyphenation points" trick
	<https://froude.eu/groff/examples/hyphenation-points.html>
	relies upon it.  Add explanatory comment.

	Fixes <https://savannah.gnu.org/bugs/?61348>.

2021-10-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* test-groff.in: Add support for sboxes.tmac.

2021-10-11  Paul Eggert <eggert@cs.ucla.edu>

	Include <config.h> before any standard headers.

	Gnulib requires that files that might use Gnulib features (which
	pretty much means every C or C++ source file) must include
	<config.h> first.  Arrange for that.  This will be needed once
	Groff updates to the latest Gnulib; see Bjarni Ingi Gislason's
	problem report in:
	<https://lists.gnu.org/r/bug-gnulib/2021-10/msg00008.html>

	The only exception I can see is src/utils/addftinfo/guess.cpp,
	which does not include any standard include file either directly
	or indirectly, and so need not include <config.h>.

	* src/devices/xditview/Dvi.c:
	* src/devices/xditview/font.c:
	* src/devices/xditview/lex.c:
	* src/devices/xditview/page.c:
	* src/devices/xditview/parse.c:
	* src/libs/libbib/map.c:
	* src/libs/libgroff/change_lf.cpp:
	* src/libs/libgroff/cmap.cpp:
	* src/libs/libgroff/cset.cpp:
	* src/libs/libgroff/fmod.c:
	* src/libs/libgroff/getcwd.c:
	* src/libs/libgroff/lf.cpp:
	* src/libs/libgroff/ptable.cpp:
	* src/libs/libgroff/quotearg.c:
	* src/libs/libxutil/DviChar.c:
	* src/libs/libxutil/XFontName.c:
	* src/libs/libxutil/xmalloc.c:
	* src/utils/indxbib/signal.c: Do it.

	[Fixes <https://savannah.gnu.org/bugs/?61315>. --GBR]

2021-10-10  Deri James  <deri@chuzzlewit.myzen.co.uk>

	Handle pdfs > v1.4 loaded by \X'pdf: pdfpic.

	* src/devices/gropdf/gropdf.pl: Improve loading of pdfs above
	version 1.4, i.e. handle compressed nodes in /ObjStm. Also improve
	code in \X'pdf: import'.

2021-10-10  Deri James  <deri@chuzzlewit.myzen.co.uk>

	Add new background boxes to gropdf.

	* src/devices/gropdf/gropdf.pl: New \X'pdf background' command.
	* tmac/pdf.tmac: Covenience command .pdfbackground added.
	* contrib/sboxes/: Files which demonstrate use of background
	boxes using -ms macros.

2021-10-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	[tests]: Fix portability problems in 2 tests.

	* src/roff/groff/tests/break_zero-length_output_line_sanely.sh:
	* tmac/tests/s_PN-works.sh: Migrate from `echo` to `printf` for
	test inputs containing backslashes to be interpreted literally.

	Fixes part of <https://savannah.gnu.org/bugs/?61302>.  Thanks to
	Deri James for identifying the problem.

2021-10-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grotty]: Slightly refactor.

	* src/devices/grotty/tty.cpp: Boolify.  Demote numerous
	variables (and one return type) from `int` to `bool`, use
	Boolean insted of integer literals with them, and give the
	variables names resembling logical predicates.
	 - horizontal_tab_flag -> want_horizontal_tabs
	 - form_feed_flag -> want_form_feeds
	 - bold_flag_option -> want_emboldening_by_overstriking
	 - bold_flag -> do_bold
	 - underline_flag_option -> want_italics_by_underlining
	 - underline_flag -> do_underline
	 - overstrike_flag -> want_glyph_composition_by_overstriking
	 - draw_flag -> allow_drawing_commands
	 - italic_flag_option -> want_sgr_italics
	 - italic_flag -> do_sgr_italics
	 - reverse_flag_option -> want_reverse_video_for_italics
	 - reverse_flag -> do_reverse_video
	 - old_drawing_scheme -> use_overstriking_drawing_scheme
	(class tty_printer:printer): Similarly.
	 - is_underline -> is_underlining
	 - is_bold -> is_boldfacing
	 - cu_flag -> is_continuously_underlining
	(tty_printer::tty_color): Demote return type as above.  Invert
	its sense; rename `unknown_color` to `is_known_color`.
	(tty_printer::color_to_idx): Invert sense of test at
	`tty_color()` call site.

2021-10-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grotty]: Slightly refactor.

	* src/devices/grotty/tty.cpp (tty_printer::special_link): Use
	consistent loop styles and drop unnecessary arithmetic.

2021-10-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Fix oversight and improve `MR` test.

	* tmac/an-ext.tmac (initialization): Enable `mY` register to
	indicate use of grotty(1) (and therefore OSC 8 support) if the
	"ascii", "cp1047", or "latin1" output devices are used, not just
	"utf8".

	* tmac/tests/an_MR-works.sh: Drop copy-and-paste cruft from a
	different test.  Update expected output and check for the
	presence of "link" device control subcommands, not just correct
	font selections.

2021-10-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Add `MR` macro for man page cross references.

	* tmac/an.tmac (an-prepare-page-title): After a possibly
	abbreviated man page title is determined, redefine `an-pageref`
	to set the title portion in the font stored in the `MF` string
	and bracket it with italic corrections if that font is thought
	to be oblique.
	(MR): Add macro to format the text of a man page cross
	reference, and hyperlink it on HTML and terminal output devices
	if permitted by the `U` register.
	(initialization): Define `MF` string as `I` if not already set.
	Define `an-lic` and `an-ic` strings as either empty or as
	containing italic corrections.

	* tmac/an-ext.tmac (MR): If the formatter is not GNU troff,
	define macro to format the text of a man page cross reference.

	* tmac/groff_man.7.man.in: Document it.
	(Description): Add macro to summary table.
	(Description/Hyperlink macros): Document new feature.  Note
	origin in Plan 9 troff.  [style] Add examples of use.
	(Description/Font style macros): Drop man page cross references
	from list of items whose typeface conventions are disputed,
	since we have a semantic macro now and a configurable means of
	resolving the problem.
	(History): Add `MR` item.
	(Options) <MF>: Document new string.
	(Files) <an-ext.tmac>: Revise discussion to accommodate `MR`.
	(Authors): Add myself an author of extension macros.

	* tmac/tests/an_MR-works.sh: Test it.
	* tmac/tmac.am (tmac_TESTS): Run test.

	* NEWS: Add item.

2021-10-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an.tmac (SH, SS): Invoke `ne` request _before_ performing
	font remapping: any page-breaking decision will be taken before
	the remapping happens, and so won't be in effect across a page
	boundary, undesirably impacting header and footer text.

	Fixes <https://savannah.gnu.org/bugs/?61279>.

2021-10-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	Regression-test Savannah #61279.

	* tmac/tests/an_font-remapping-does-not-affect-titles.sh: Do it.
	* tmac/tmac.am (tmac_TESTS): Run test.

2021-10-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an.tmac (an-abbreviate-title): Rename to...
	(an-prepare-page-title): ...this.

2021-10-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	[docs]: Correct erroneous claim: the ".NH S" extension to ms(7)
	comes from 4.2BSD, not GNU.

2021-10-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/ms.ms (Creating a table of contents): Sync with recent
	additions to groff_ms(7), expand, and heavily revise.

2021-10-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	[andoc,man,mdoc]: Fix Savannah #61266.  Resolve problems in
	batch rendering of man pages to PDF arising from entanglement
	of end-of-input traps, page location traps, continuous rendering
	mode, and andoc's reloading of the (m)an and (m)doc packages.

	* tmac/andoc.tmac (reload-doc, reload-man): Remove end-of-input
	traps alongside others.

	* tmac/an.tmac (an-end): Only perform flush and "manual" page
	footer placement if in continuous rendering mode, since this
	macro is not only called by a trap placed only in continuous
	rendering mode, but also by andoc when changing macro packages.
	Unconditionally remove the `an-header` trap since the next
	document might be using a different macro package.

	* tmac/mdoc/doc-common: In initialization, set flag indicating
	that manual header placement will be required.
	(Dt): Call `doc-setup-header` (which sets up several types of
	trap) unconditionally, and break the page if the vertical
	drawing position is anywhere but at the top.
	(Os): If the package has just been initialized, call
	`doc-header` to force the page header to be written.
	(doc-end-macro): Remove `doc-header` trap since the next
	document might be using a different macro package.  Break the
	page.  Set flag indicating that manual header placement will be
	required for the next document.

	* tmac/mdoc/doc-ditroff (doc-setup-header): Only set page
	location traps for the header and footer if not continuously
	rendering.
	* tmac/mdoc/doc-nroff (doc-setup-header): Stop calling
	`doc-header` here if continuously rendering.  Emit an
	unconditional break.  Except for the location of the footer
	trap, the `doc-setup-header` implementations are now identical.

	Refactoring is needed: some macros and registers have misleading
	names, there is some code duplication in mdoc, and some of the
	trap management problems are solved in slightly different ways
	in man(7) and mdoc(7), perhaps unnecessarily.  We also need some
	test scripts to protect us from regressions.  But this fixes the
	rendering problems.

	Fixes <https://savannah.gnu.org/bugs/?61266>.

2021-10-02  Keith Marshall  <keith.d.marshall@ntlworld.com>

	Correct manpage typo/inconsistency.

	* tmac/groff_ms.7.man (TC_LEADER, TC_MARGIN): Correct; should
	be...
	(TC\-LEADER, TC\-MARGIN): ...respectively, for consistency
	with...
	* tmac/s.tmac (TC-LEADER, TC-MARGIN): ...these.

2021-10-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (encode_char): Update diagnostic
	messages to not presume the identity of the escape character;
	continues 9be3f8e3 (4 June).

2021-10-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/devices/grotty/tty.cpp (tty_printer:special_link): Tighten
	code slightly by taking advantage of preprocessor string literal
	concatenation.

2021-10-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Fix `U` enablement override so it actually works the way
	I had in mind.
	- The package proper enables it by default, but
	- the sample site configuration file switches it off, on the
	  assumption that distributors and users will know better than
	  we do when they get a capable pager in place, and
	- a register setting at the command line (or otherwise prior to
	  the loading of the macro package) is dispositive.

	* tmac/an-ext.tmac: Rename URI output device capability register
	from `U` to `mU`.
	(UR, UE, MT, ME): Test both `U` and `mU` before handling
	hyperlinks specially.
	* tmac/an.tmac: If `man.local` did not initialize `U`, switch it
	on.  The hope is that we can drop the stuff in `man.local`
	completely at some point.
	* tman/man.local: Only initialize `U` if it is not already
	defined.

2021-10-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Use OSC 8 hyperlinks.

	* tmac/an-ext.tmac: Add registers `mY` (output driver is
	"grotty") and `U` (generate hyperlinks).  `U` is intended to
	apply to hyperlinking support in general, not just grotty's OSC
	8 feature.  If the output driver is either grohtml or grotty,
	make `U` true (`man.local` is read subsequently and can override
	it).
	(UR, MT): Conditionalize diversion production on hyperlink
	support, not grohtml use per se.
	(UE, ME): Generate device control commands for `tty` device to
	emit hyperlinks.

	* tmac/groff_man.7.man.in (Options): Document -rU.
	(Files) [style]: Add example of `U` register enablement.

	* tmac/man.local: Disable `U` register by default, except on the
	`html` device.

	* NEWS: Add item.

2021-10-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grotty]: Add OSC 8 hyperlink support.

	* src/devices/grotty/tty.cpp: Do it.  Define `OSC` (Operating
	System Command) and `ST` (String Terminator) preprocessor
	symbols for these ECMA-48 (ISO 6429) character sequences.
	(tty_printer::simple_add_char): Add stripped-down alternative to
	`add_char()` member function for cases where we want to use many
	defaults because we're writing a terminal escape sequence, not
	rendering a glyph.  (A function like `add_char()` that takes 8
	arguments of varying types is a code smell--phew!)
	(tty_printer::special): Call `special_link()` member function if
	the device control command is `link`.
	(tty_printer::special_link): Add new member function to generate
	OSC 8 hyperlinks.

	* src/devices/grotty/grotty.1.man: Document it.  Observe in
	multiple places that disablement of SGR escape sequences
	disables OSC 8 too.

	* src/devices/grotty/tests/osc8_works.sh: Test it.
	* src/devices/grotty/grotty.am (TESTS, grotty_TESTS): Run test.
	(EXTRA_DIST): Ship test.

	* NEWS: Add item.

	Fixes <https://savannah.gnu.org/bugs/?60666>.  Thanks to Steffen
	Nurpmeso for supplying a proof-of-concept.  (I went with my own
	implementation, though, so blame me if it breaks.)

2021-10-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Convert special character glyphs corresponding to
	Unicode Basic Latin ("ASCII") code points to those code points
	when they occur in device escapes.  (They should be correct for
	IBM code page 1047 as well, but this is untested.)  This is
	necessary for encoding URLs in device control commands.  Special
	character identifiers are presumed to be the defaults documented
	in groff_char(7); this is a design gap that we should consider
	addressing.  (We don't have a way to ask "is this the special
	character corresponding to Unicode basic Latin code point X?")

	* src/roff/troff/input.cpp (encode_char): Do it.

2021-09-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man, mdoc]: Draw line after each page more consistently in
	continuous rendering mode.

	* tmac/an.tmac (an-bp): Move responsibility for line-drawing
	after each document from here...
	(an-end): ...to here.  Refactor; drop
	`an-do-draw-line-after-document` register in favor of testing
	`\n[.F]` for nullity; if it is not null, then another document
	is due to be processed (in fact, `.F` already contains its file
	name), and we need to draw a separating line.  If it is null, we
	have reached the end of input and none is necessary.

	* tmac/mdoc/doc-ditroff.tmac (doc-setup-page-layout):
	* tmac/mdoc/doc-nroff.tmac (doc-setup-page-layout): Save
	configured line length in new register `doc-line-length`.
	* tmac/mdoc/doc-common.tmac (doc-end-macro): Port the above
	man(7) feature to mdoc(7); draw a line of this length after each
	document if continuously rendering.

2021-10-01  Keith Marshall  <keith.d.marshall@ntlworld.com>

	[ms]: Update footnote handling documentation.

	* tmac/groff_ms.7.man: Make some linguistic style adjustments.
	(FP): Document it; see <https://savannah.gnu.org/bugs/?61022>.
	(FS-MARK): Likewise, document this recently added feature.

2021-10-01  Keith Marshall  <keith.d.marshall@ntlworld.com>

	[ms]: Support user-defined TOC leader style.

	* tmac/s.tmac (toc*leader-char): Rename it as...
	(TC-LEADER): ...this; within PX, leave it unchanged, if
	predefined.
	(TC-MARGIN): New numeric register; if predefined, leave it
	unchanged, otherwise define as \w'000'; use it instead of
	hard-wired definition, within PX; see
	<https://savannah.gnu.org/bugs/?61157>.

	* tmac/groff_ms.7.man (TC-LEADER, TC-MARGIN): Document them.

2021-09-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	[pic]: Update input file name correctly.

	* src/preproc/pic/troff.cpp (troff_output::set_location): Copy
	the inbound file name argument with `strdup()` and store pointer
	to this copy in `last_filename` instead of aliasing argument.
	(troff_output::~troff_output): `free()` the memory allocated by
	`strdup()` in destructor.

	Fixes <https://savannah.gnu.org/bugs/?61206>.  Pointer
	assignment instead of a string copy was being performed, which
	means that after its initial assignment, `last_filename` always
	shared its value ultimately with the `current_filename` pointer
	{a libgroff symbol}, so `strcmp()` was always being performed on
	identical pointers.  Problem appears to date back 30+ years, to
	the dawn of our repo history.

2021-09-24  G. Branden Robinson <g.branden.robinson@gmail.com>

	* font/devlj4/S: Make font name consistent with file name.

2021-09-24  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libgroff/font.cpp (font::load): Throw error if a font
	description file is missing a `spacewidth` directive, since it
	is now re-documented as mandatory.  Atypically, we don't return
	false in this scenario; instead we proceed with the existing
	logic to compute a space width for the font based on typical
	practices for Western alphabetic fonts (see, e.g.,
	<https://type.today/en/journal/spaces>.

	* font/devdvi/EX:
	* font/devdvi/MI:
	* font/devdvi/S:
	* font/devdvi/SA:
	* font/devps/EURO: Add `spacewidth` directive to font
	description files we ship that were missing it.  We use the same
	values that libgroff would have computed, so there should be no
	visible difference.

2021-09-24  G. Branden Robinson <g.branden.robinson@gmail.com>

	* Makefile.am: Mark `$(TEST_SUITE_LOG)` as `.PRECIOUS`.  This
	prevents the test suite log from being deleted if there are any
	failures.

	Fixes <https://savannah.gnu.org/bugs/?61213>.  Thanks to Bjarni
	Ingi Gislason for the report.

2021-09-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.pl (do_line): Move first-macro-call test
	for `TH` to a more appropriate location.

2021-09-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grog]: Refactor to make it easier to generalize a per-package
	scoring mechanism.

	* src/utils/grog/grog.pl: Convert `inferred_main_package` from
	scalar to list.  Add scalar `selected_main_package` to store a
	"winner" from the list of main packages inferred.
	(push_main_package): Add subroutine to populate
	`inferred_main_package`.
	(do_line, infer_man_or_ms_package): Update to use
	`push_main_package`.
	(construct_command): Rewrite to select a main package from the
	list `inferred_main_package`, taking user-supplied arguments
	into account as an overriding factor.  Drop now-unused `msupp`
	list.

2021-09-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/groff.cpp (main): If the `-I` option is given,
	run the grn preprocessor and pass it an `-M` option with `-I`'s
	argument.

	* NEWS:
	* src/roff/groff/groff.1.man (Options): Document it.  Also
	document that `-I` options are passed to the output driver.

2021-09-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.pl (do_line): Don't strip groff-style
	comments \# if AT&T compatibility mode is enabled.

2021-09-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Perform more font description file validation.  Our
	documentation said that the `name` directive was mandatory but
	we didn't actually enforce this.  Also be more specific in our
	complaints of ill-formedness.

	* src/libs/libgroff/font.cpp (font::load): Add local Boolean
	variable `saw_name_directive`.  Rename `had_charset` to
	`saw_charset_directive` for symmetry.  Emit distinct diagnostics
	for `spacewidth` and `slant` directives having a missing,
	non-numeric, or out-of-range argument.  Emit diagnostic if the
	file lacks a `name` directive.

2021-09-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Slightly refactor.

	* src/libs/libgroff/font.cpp (struct text_file): Rename member
	variable `size` to `linebufsize` for clarity (it's certainly not
	the size of the text file).
	(text_file::text_file): Initialize `linebufsize` in constructor.
	(text_file::next_line): Allocate new `buf` using `linebufsize`
	instead of magic number (which was also stored to `size`, making
	it only semi-magic, I guess).  Rename local variable `i` to
	`length` since it is in fact the length (in `char`s) of the
	string stored in `buf`.

2021-09-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Revisit fix for Savannah #61173.

	* src/libs/libgroff/font.cpp (text_file::text_file): Restore
	setting of `lineno` member variable to `0` in constructor
	initializer list, but...
	(text_file::next_line): ...unconditionally increment it every
	time through this member function's outer loop.

2021-09-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Align diagnostic message format with libgroff.

	* src/roff/troff/input.cpp (fatal_with_file_and_line): Report
	the program name if one is defined.
	(fatal_with_file_and_line, error_with_file_and_line,
	debug_with_file_and_line): Report the line number of the input
	file only if it is positive, so that other values can be used
	for "whole file" complaints.  See commit cd0457b6, 17 September.

2021-09-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Drop redundant diagnostic.

	* src/libs/libgroff/font.cpp (font::load): Drop redundant
	diagnostic; a missing `charset` directive is already checked for
	later, circa line 998.

2021-09-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Fix code style nits.

	* src/libs/libgroff/font.cpp (font::font): Use initializer list
	as much as possible in constructor.
	(font::unit_scale, font::get_width, font::get_height,
	font::get_depth, font::get_italic_correction,
	font::get_left_italic_correction,
	font::get_subscript_correction, font::get_character_type,
	font::get_code, font::get_special_device_encoding): These member
	functions rely on the font being either indexed or associated
	with an output device that uses the `unicode` directive in its
	`DESC` file.  The compiler doesn't know this, so annotate the
	`abort()` calls that conclude them.  Add an `assert()` before
	each one to more usefully describe the violated invariant in the
	event this unreachable code is somehow reached.
	(font::get_width): Annotate special meaning of zero zoom factor.

2021-09-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/include/font.h (UNDEFINED_GLYPH): Use idiomatic C++98 null
	pointer constant.

2021-09-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grohtml]: Fix compiler warning and dead store.

	* src/devices/grohtml/post-html.cpp (html_printer::special):
	Update call of `font::load_font()` to stop passing obsolete
	parameter (deleted by me in 2dff87d3, 17 September).  The value
	stored to the parameter by that function was never read anyway,
	so also get rid of the variable that contained it.

	Fixes <https://savannah.gnu.org/bugs/?61187>.  Thanks to Bjarni
	Ingi Gislason for the report.

2021-09-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Slightly refactor.

	* src/include/font.h (font::scan_papersize): Demote return type
	from `int` to `bool`.
	* src/libs/libgroff/font.cpp (font::scan_papersize): As above.
	Use Boolean instead of integer literals.  Also rename `FILE`
	stream pointer from `f` to `fp` and perform an explicit
	comparison against the idiomatic C++98 null pointer constant.

2021-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff, troff]: Slightly refactor device and font description
	file loading.  Remove dead code.

	* src/include/font.h (font::load_font): Drop second parameter.
	It was never used for its intended purpose.
	(load): Drop first parameter; likewise.
	* src/libs/libgroff/font.cpp (font::load, load): As above.

	* src/libs/libgroff/font.cpp (struct text_file): Rename
	`skip_comments` to `recognize_comments`.  Demote that and
	`silent` from `int` to `bool`.
	(text_file::text_file): Use Boolean rather than integer literals
	in constructor.
	(text_file::next_line, font::load): Apply above renaming.

	* src/libs/libgroff/font.cpp (font::load): Rename parameter from
	`head_only` to `load_header_only` to be more communicative.
	Drop test of font description file name being `DESC`; this code
	was not being reached.  Stop throwing errors from this function
	on failure to open the file; the caller will handle this when it
	sees our false return value.  Rename local variable `command` to
	`directive` for alignment with our documentation.  Replace "I
	dont think this should happen" test and comment with `assert()`.
	(struct table): Rename member from `command` to
	`numeric_directive` to indicate its specificity, tracking only a
	subset of valid `DESC` file directives.

	* src/libs/libgroff/font.cpp (font::load, font::load_desc):
	Remove redundant assignments to the member variable formerly
	known as `skip_comments`.

	* src/libs/libgroff/font.cpp (font::load_desc): Rename local
	variable `directive_found` to `numeric_directive_found` to
	clarify logic.

	* src/roff/troff/node.cpp (mount_font_no_translate): Simplify
	call of `font::load_font`.  The `not_found` in-out parameter
	which was so agonizingly passed up through layers of library
	calls was never actually read.  Drop code that has been `#if
	0`-ed out since 1993.

2021-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libgroff/font.cpp (font::load_desc): Clear line
	number before emitting whole-file validity diagnostics.

2021-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libgroff/font.cpp (font::load_desc): Emit correct
	line numbers when complaining of invalid `sizescale`, `hor`, or
	`vert` values in device description files.

	Fixes <https://savannah.gnu.org/bugs/?61174>.

2021-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Increase validation of device and font description
	files.

	* src/libs/libgroff/font.cpp (font::load): Validate the syntax
	and value of the `name` directive.
	(font::load_desc): Issue distinct diagnostics for a `fonts`
	directive that is missing arguments and for a first argument
	that can't be interpreted as a valid number.

2021-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Make error reporting more flexible.

	* src/libs/libgroff/error.cpp (do_error_with_file_and_line):
	Interpret a nonpositive line number specially: treat the
	diagnostic as applying to the entire file, and omit the line
	number.

2021-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Add information to diagnostics.

	* src/libs/libgroff/font.cpp (font::load): Drop unused argument
	from `error()` call.
	(font::load_desc): In another, add name of numeric directive
	that is given an unparsable number.

2021-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Tweak code style.

	* src/libs/libgroff/font.cpp (trim_args, font::load,
	font::load_desc): Compare pointer explicitly to null pointer
	literal `0` instead of using logical complementation.
	(font::load, font::load_desc): Swap order of null pointer
	equality comparisons when a typo or thinko could lead to lvalue
	assignment.

2021-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Use idiomatic C++98 null pointer constant.

	* src/libs/libgroff/font.cpp (text_file::error,
	glyph_to_unicode, font::get_special_device_encoding,
	font::load):
	* src/libs/libgroff/fontfile.cpp (font::image_generator):
	* src/libs/libgroff/nametoindex.cpp (class charinfo,
	character_indexer::named_char_glyph,
	character_indexer::numbered_char_glyph):
	* src/libs/libgroff/string.cpp (string::clear):
	* src/libs/libgroff/tmpfile.cpp (temp_init::temp_init):
	* src/libs/libgroff/tmpname.cpp (gen_tempname): Use `0` literal
	instead of `NULL` to represent a null pointer; this was
	idiomatic in C++98 and is mostly the practice elsewhere in
	groff.  Also swap order of null pointer equality comparisons
	when a typo or thinko could lead to lvalue assignment.

	* src/roff/troff/input.cpp (input_stack::diversion_state,
	input_stack::get_diversion_state, charinfo::contains,
	glyph_to_name): Similar.

2021-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Slightly refactor.

	* src/libs/libgroff/font.cpp (font::load): Use same loop style
	as `font::load_desc()`.

2021-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Fix off-by-one error in font and device description
	file reader diagnostics.

	* src/libs/libgroff/font.cpp (text_file::text_file): All text
	files begin with line 1, not line 0.

	Fixes <https://savannah.gnu.org/bugs/?61173>.

2021-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Boolify more interfaces.

	* src/include/font.h (font::unit_scale):
	* src/libs/libgroff/font.cpp (font::unit_scale): Demote return
	type from `int` to `bool`.  Use Boolean literals instead of
	integers.

	* src/libs/libgroff/font.cpp (struct text_file): Rename `next`
	member function to `next_line`.  Demote its return type from
	`int` to `bool`.  Use Boolean literals instead of integers.

2021-09-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Boolify local variables.

	* src/libs/libgroff/font.cpp (font::load, font::load_desc):
	Demote local variables used as Booleans from `int` to `bool` and
	update literals used with them from integer to Boolean.

2021-09-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grn, groff, troff]: Communicate better when device DESC file
	can't be loaded.

	* src/libs/libgroff/font.cpp (font::load_desc): Stop writing
	diagnostic to standard error.
	* src/preproc/grn/main.cpp (getres):
	* src/roff/troff/troff.cpp (main): Replace "sorry, I can't
	continue" fatal diagnostic with a more informative message.
	Instead say which device's DESC file the program was trying to
	open, which might reveal a user's logic error or typo.
	* src/roff/groff/groff.cpp (main): Replace "invalid device" text
	of fatal diagnostic similarly.  I think it is more helpful to
	indicate the operation that failed rather than saying the device
	was invalid, particularly since there are other ways for a
	device description to be invalid even if a DESC file is found
	and loaded.

2021-09-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Throw warning diagnostic if device DESC file attempts
	to load unavailable fonts.

	* src/roff/troff/input.cpp (main): Take advantage of new Boolean
	return values of `mount_style()` and `mount_font()` to produce
	warning diagnostics if mounting a style or font as directed by
	the DESC file fails.  Explain in a comment why, at present, this
	style mount warning will never actually trip.

2021-09-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Lift font mounting diagnostic messages to be closer to
	their user-controlled contexts to provide more information.  In
	many cases no diagnostic was being thrown at all when an
	unavailable font was requested by name, which is the method most
	users prefer, and which meant that failures resulting from typos
	in font names for many requests (`uf`, `fschar`, `rfschar`,
	`special`, `fspecial`, `fzoom`, `bd`, `tkf`, `cs`) were going
	unreported.  Further, promote these font warnings to errors
	because the request will utterly fail to do what was requested
	with no reasonable fallback.  Possibly, they were formerly
	warnings because at the low level they were being emitted, they
	could also have originated from unavailable fonts encountered in
	device description files, and while that's bad news, it results
	in no formatting problems if it doesn't affect fonts that an
	input document actually uses.

	* src/roff/troff/node.cpp (struct font_lookup_info): New struct
	keeps the font name or position requested, and the position of
	successful font lookup.
	(font_lookup_info::font_lookup_info): Add constructor.
	(font_lookup_error): New function builds error message using a
	`font_lookup_info` struct and a message argument.
	(get_fontno): Rename to...
	(has_font): ...this.  Add argument to take a pointer to a
	`font_lookup_info` struct.  Return a `bool` indicating whether
	the lookup succeeded.  Place former `int` return value into the
	struct instead.  Populate the other struct members with the
	requested font name or position, as appropriate.
	(mount_font_no_translate): Stop throwing warning diagnostic here
	if a font cannot be loaded.  Instead, throw them...
	(font_position): ...here, and...
	(underline_font, define_font_special_character,
	remove_font_special_character, read_special_fonts,
	font_special_request, font_zoom_request, bold_font, track_kern,
	constant_space): ...here, using `font_lookup_info` structs and
	`has_font()`.
	(remove_font_special_character): Stop returning early if font
	lookup fails; it's gratuitously inconsistent with other similar
	functions (save one, which has a reason to be different).
	(define_font_special_character): Return early if font lookup
	fails and say why in a comment (we can't `skip_line()`).

2021-09-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	* [libgroff, troff]: Further boolify.

	* src/include/font.h (load_font, font): Demote parameters from
	{pointer to} `int` to `bool` and update default literal from
	integer to Boolean (except for the pointer).  Update comment.
	* src/libs/libgroff/font.cpp (load_font, load): Similarly.

	* src/roff/troff/node.h (mount_font):
	* src/roff/troff/node.cpp (mount_font): Demote return type from
	`int` to `bool`.

	* src/roff/troff/node.h (mount_style):
	* src/roff/troff/node.cpp (mount_style): Promote return type
	from `void` to `bool`.

	* src/roff/troff/node.cpp (mount_font_no_translate): Demote
	return type and `check_only` parameter from `int` to `bool` and
	use Boolean rather than integer literals with them.
	(check_font): Update call site of `mount_font_no_translate` to
	use Boolean literal.
	(font_position): Indicate that "error" is ignored by casting
	return value of `mount_font` to void instead of using a comment.
	(style): Cast return value of `mount_style` to `void`.

2021-09-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	* Makefile.am: Add `.DELETE_ON_ERROR` special target; both GNU
	and FreeBSD make(1)s support it so maybe it will be portable
	enough.  This avoids, among other problems, a target appearing
	falsely up-to-date (often as an empty file) when the troff
	process generating it experiences an assertion failure.

2021-09-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	Since June 1991 if not earlier, groff (technically, the refer,
	lookbib, and lkbib programs) has trusted the header contents of
	binary bibliographic index files (canonically generated with
	indxbib(1)) regarding the sizes of the data structures that
	follow in the file, a notorious class of security problem.  In
	July 2013, the Mayhem Team at Carnegie Mellon University
	reported to the Debian Bug Tracking System a problem with
	groff's refer(1) implementation dumping core when reading an
	index file prepared by a fuzzer.

	* src/libs/libbib/index.cpp (index_search_item::check_header):
	Add new member function to validate the header of an indexed
	bibliography file, returning a string describing in detail the
	first validity problem encountered.
	(index_search_item::load): Perform the foregoing check before
	using any of the size values taken from the header; they are
	relied upon for pointer arithmetic.  If in verification mode
	{using the undocumented `-V` flag to refer(1), lkbib(1), or
	lookbib(1)}, report the details of the problem encountered.
	Regardless of that mode, if there is a validity problem, report
	corruption of the index file and abandon it, forcing fallback to
	the text version of the corresponding bibliography file.

	Fixes <https://bugs.debian.org/716109>.

2021-09-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libbib/index.cpp
	(index_search_item::get_invalidity_reason): Don't complain about
	a last list element being nonnegative if the list size was zero
	in the first place, as can happen with an empty index.  More
	seriously from a language standpoint, avoid reading through a
	negative array index (this can escape a compiler's attention
	because we're reading from the midst of a heap-allocated or
	`mmap()`ed region, but it's still a code smell).

	Fixes <https://savannah.gnu.org/bugs/?61144>.

2021-09-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libbib/index.cpp (index_search_item::load,
	index_search_item::get_invalidity_reason,
	index_search_item::add_out_of_date_file): Clarify diagnostic
	messages.  Make it more obvious when we're complaining of a
	problem in an indexed bibliographic database file (rather than a
	plain text one).  Dial down the amount of Unix jargon a little.

2021-09-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libbib]: Partially boolify.

	* src/include/search.h:
	* src/libs/libbib/index.cpp:
	* src/preproc/refer/refer.cpp:
	* src/utils/lkbib/lkbib.cpp:
	* src/utils/lookbib/lookbib.cpp: Demote `verify_flag` from an
	`int` to a `bool`, rename it to `do_verify`, use Boolean
	instead of integer literals with it, and update call sites.

	* src/libs/libbib/index.cpp: Rename `do_verify()` member
	function to `get_invalidity_reason()` (it returns a string).
	Demote `load()` from `int` to `bool`; do the same for `verify()`
	and rename it to `is_valid()` as well.  Use Boolean instead of
	integer literals with them.
	(index_search_item::is_valid, make_index_search_item): Update
	call sites of renamed member functions.

2021-09-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libbib/map.c (mapread): Compare return value of
	`mmap()` to `MAP_FAILED` instead of `(char *)-1`.  `MAP_FAILED`
	is documented in POSIX Issue 5 a.k.a. SUSv2
	<https://pubs.opengroup.org/onlinepubs/007908799/xsh/mmap.html>
	{1997} and should be portable enough by now.

2021-09-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Clamp negative tab stop positions to zero instead of
	throwing an assertion failure.

	* src/roff/troff/env.cpp (tab_stops::distance_to_next_tab):
	Replace `assert` with clamping logic, ensuring that `lastpos`
	can never be negative.  While negative tab stop positions don't
	make much sense (they result in zero horizontal motion), user
	input like `.ta T -5` should never provoke an assertion failure.

	(set_tabs): Throw range warning in additional scenario, viz., if
	a repeating tab offset is negative.

	Fixes <https://bugs.debian.org/990406>.  Thanks to наб for the
	report.

2021-09-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Boolify `set_tabs` function.

	* src/roff/troff/env.cpp (set_tabs): Demote local variables from
	`int` to `bool` and give them predicate-like names.
	  - `first` -> `is_first_stop`
	  - `repeated` -> `is_repeating_stop`
	Use Boolean instead of integer literals in assignments to them.

2021-09-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Deprecate `OP` macro.

	* tmac/an-ext.tmac (OP): Move implementation from here...
	* tmac/an.tmac (OP): ...to here.  Throw deprecation warning.
	Throw style warning if wrong number of arguments given.

	* tmac/groff_man.7.man.in (Description): Drop from introductory
	macro summary.
	(Description/Command synopsis macros): Move discussion from
	here...
	(Description/Deprecated features): ...to here.  Explain why it's
	deprecated.

2021-09-07  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Make `ab` request quiet if given no arguments.  The
	`tm`, `tm1`, and `tmc` requests provide ample flexibility for
	constructing diagnostic messages prior to an error exit, and
	it's convenient to make `ab` itself quiet instead of saying
	"User Abort." as Unix Version 7 troff did.  Further, there is no
	standardization across troffs regarding what `ab` should emit if
	arguments are absent.

	* src/roff/troff/input.cpp (abort_request): Do it.

	* src/roff/groff/tests/ab_works.sh:
	* src/roff/groff/groff.am (groff_TESTS): Test it.

	* doc/groff.texi (Debugging, Implementation Differences):
	* man/groff.7.man (Requests/Request short reference):
	* man/groff_diff.7.man (Implementation differences): Document
	it.

	Fixes <https://savannah.gnu.org/bugs/?61070>.  Thanks to Dave
	Kemper for the discussion.

2021-09-07  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Update exit status literals.

	* src/roff/troff/div.cpp (top_level_diversion::begin_page):
	* src/roff/troff/input.cpp (exit_troff, abort_request, do_error,
	fatal_with_file_and_line): Use standard C library preprocessor
	symbols `EXIT_SUCCESS` and `EXIT_FAILURE` instead of 0 and 1
	literals, respectively.

2021-09-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Drop unused member function.

	* src/roff/troff/token.h (class token): Drop unused `title`
	member function.  Its implementation was apparently removed in
	the prehistory of our Git repository: see James Clark's
	ChangeLog entry of 1990-09-06.

2021-09-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Remove workaround for Cfront 1.2 bug.

	* src/roff/troff/input.cpp (do_define_macro): Move `dot_symbol`
	from to file to function local scope as originally intended.

2021-09-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Rename parameter.

	* src/roff/troff/input.cpp (token::usable_as_delimiter): Rename
	parameter from `err` to `report_error` to be more communicative.

2021-09-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Boolify members of `token` class.

	* src/roff/troff/token.h (class token): Demote return type of
	several member functions from `int` to `bool` and rename them.
	  - `backspace` -> `is_backspace`
	  - `delimiter` -> `usable_as_delimiter`
	  - `dummy` -> `is_dummy`
	  - `eof` -> `is_eof`
	  - `horizontal_space` -> `is_horizontal_space`
	  - `hyphen_indicator` -> `is_hyphen_indicator`
	  - `leader` -> `is_leader`
	  - `left_brace` -> `is_left_brace`
	  - `newline` -> `is_newline`
	  - `page_ejector` -> `is_page_ejector`
	  - `right_brace` -> `is_right_brace`
	  - `space` -> `is_space`
	  - `special` -> `is_special`
	  - `stretchable_space` -> `is_stretchable_space`
	  - `tab` -> `is_tab`
	  - `transparent_dummy` -> `is_transparent_dummy`
	  - `transparent` -> `is_transparent`
	  - `unstretchable_space` -> `is_unstretchable_space`
	  - `white_space` -> `is_white_space`
	  - `zero_width_break` -> `is_zero_width_break`
	(class token): Drop 1991 comment anticipating that member
	function `nspaces` could return "2" for a "double space"; this
	was apparently never implemented.
	(class token): Drop parameter names from declarations; the
	prevailing style (familiar from Stroustrup) is not to use them.
	(token::is_special): Simplify implementation.

	* src/roff/troff/input.cpp (has_arg,
	token::usable_as_delimiter): Update definitions of above member
	functions not defined in token.h.

	* src/roff/troff/div.cpp (begin_page, space_request, need_space,
	output_saved_vertical_space, flush_output):
	* src/roff/troff/env.cpp (fill, no_fill, center,
	right_justify, indent, temporary_indent, margin_character,
	number_lines, do_break_request, hyphen_word):
	* src/roff/troff/input.cpp (next_file, do_overstrike,
	do_bracket, do_name_test, do_expr_test, do_zero_width,
	token::skip, has_arg, skip_line, empty_name_warning,
	non_empty_name_warning, do_get_long_name, process_input_stack,
	flush_pending_lines, decode_args, read_request,
	do_define_string, do_define_character, remove_character,
	do_define_macro, length_request, get_delim_number, get_line_arg,
	read_size, get_delim_name, do_register, do_width,
	read_title_parts, encode_char, do_special, device_request,
	output_request, skip_alternative, begin_alternative,
	nop_request, do_if_request, do_source, pipe_source,
	ps_bbox_request, tag, taga, do_terminal, do_translate,
	hyphenation_code, hyphenation_patterns_file_code, define_class,
	get_optional_char, check_missing_character, abort_request,
	copy_file, transparent_file, do_macro_source,
	charinfo_to_node_list, read_draw_node, read_color_draw_node):
	* src/roff/troff/node.cpp (get_fontno,
	remove_font_special_character, bold_font):
	* src/roff/troff/number.cpp (start_number, parse_term):
	* src/roff/troff/reg.cpp (define_number_reg, alter_format):
	Update call sites to use new names.

	* src/roff/troff/input.cpp (token::usable_as_delimiter,
	read_draw_node): Return Boolean, not integer, literals.

2021-09-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Clamp line and title lengths to device horizontal
	resolution.

	* src/roff/troff/env.cpp (line_length, title_length): Do it.

	Fixes <https://savannah.gnu.org/bugs/?61116>.

2021-09-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/env.cpp (do_hyphenation_patterns_file):
	Refactor slightly.  Demote `append` from `int` to `bool`.
	(hyphenation_patterns_file, hyphenation_patterns_file_append):
	Update call sites to use Boolean, not integer, literals.

2021-09-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/hvunits.h:
	* src/roff/groff/number.cpp: Mark `H0`, `V0` objects as `const`.

2021-09-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/env.cpp (do_underline_special): Refactor
	slightly.
	  - Rename `underline_spaces` -> `do_underline_spaces`.
	  - Demote it from an `int` to a `bool`.
	  - Use ternary operator and explicit character literals when
	    writing device control command instead of doing arithmetic
	    on a character literal.
	  - Update call sites to use Boolean literals.
	  - Relocate function to avoid forward reference.
	  - Mark function as static to eliminate external linkage.
	    Remove now-unnecessary prototype.

2021-09-04  Keith Marshall  <keith.d.marshall@ntlworld.com>

	Add "ms" footnote marker placement hook.

	* tmac/s.tmac (FS-MARK): New macro; define as no-op, by default.
	(@FS): Invoke FS-MARK as first action, allowing caller to "hook" it.

2021-09-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/env.cpp (distribute_space): Refactor slightly.
	  - Rename `force_reverse` to `force_reverse_node_list`.
	  - Rename `reverse` to `do_reverse`.
	  - Demote both of the above from `int` to `bool`.
	  - Use Boolean literals with them.
	  - Add assertions to enforce positive values of `nspaces` and
	    `desired_space`.
	  - Remove now-redundant test for `nspaces` being positive.
	  - Add explanatory comments.
	(environment::wrap_up_field): Update call sites of
	`distribute_space` when non-default value of
	`force_reverse_node_list` is supplied.

2021-09-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Don't adjust nonadjustable lines.  This means that the
	direction from which an output line in adjustment mode "b" (or
	its "n" synonym) is filled with supplemental space is not
	changed if that output line does not require adjustment.  This
	will result in whitespace changes to documents using that
	adjustment mode, and these changes will be plainly visible on
	low-resolution output devices like terminals.

	To illustrate, in the following "A" means an output line
	requiring adjustment; "F" a line that is "full" and does not;
	and "L" and "R" indicate distribution of adjustment spaces from
	the left and right, respectively.

	groff 1.22.4    groff 1.23.0
	------------    ------------
	A    L          A    L
	A    R          A    R
	F    L          F    R
	A    R          A    L

	* src/roff/troff/env.cpp (distribute_space): Return early if
	either the amount of desired space to be distributed or the
	count of space nodes in the output line to distribute it among
	is zero.

	* tmac/tests/an_TH-repairs-ad-damage.sh: Update test to expect
	space to be distributed differently.

	Fixes <https://savannah.gnu.org/bugs/?61089> and
	<https://savannah.gnu.org/bugs/?60673>.

2021-09-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	Skip core-dump-checking test if a core file already exists.

	* src/roff/groff/tests/regression_savannah_59202.sh: Skip test
	if a core dump is already present.

2021-09-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #60189.

	* src/roff/groff/tests/break_zero-length_output_line_sanely.sh:
	Do it.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

2021-08-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Tweak diagnostic message.

	* src/roff/troff/env.cpp (environment::environment): Use
	terminology more rigorously in diagnostic.

2021-08-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Fix code style nit.

	* src/libs/libgroff/errarg.cpp (errprint): Replace `assert(0)`
	with a meaningful predicate.

2021-08-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Demote `need_space` to Boolean.

	* src/libs/libgroff/error.cpp (do_error_with_file_and_line):
	Demote local integer `need_space` and its assignments to `bool`.

2021-08-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Boolify `font::load()` and `font::load_desc()`
	return values.

	* src/include/font.h (font::load, font::load_desc):
	* src/libs/libgroff/font.cpp (font::load, font::load_desc): Do
	it.

2021-08-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	* m4/groff.m4 (TRADITIONAL_CPP): Update `AC_MSG_CHECKING`
	argument to refer to "pre-ISO C90 syntax" instead of a
	"traditional" preprocessor.  Call `AC_MSG_ERROR` (aborting
	configuration) if the check (for pre-C90 transformation)
	succeeds.  We thus withdraw support for such superannuated
	{"Reiser"} C preprocessors.

2021-08-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Un-indirect token concatenation through macro.

	* src/include/itable.h:
	* src/include/ptable.h: Do it.

2021-08-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Drop support for `TRADITIONAL_CPP`.  This means a C
	preprocessor that does not support the ANSI C89/ISO C90
	token concatenation operator "##".

	* src/include/itable.h:
	* src/include/ptable.h: Do it.

	* Makefile.am: Undocument preprocessor symbol.

2021-08-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Drop `a_delete` preprocessor wrapper for the
	`delete` operator.  It has been rendered unnecessary by the
	removal of support for ancient C++ compilers.

	* src/include/lib.h (a_delete): Drop symbol definition.

	* src/devices/grohtml/output.cpp (word::~word):
	* src/devices/grohtml/post-html.cpp (char_block::char_block,
	assert_state::~assert_state, assert_state::add,
	assert_state::close, replace_negate_str):
	* src/devices/grops/ps.cpp (ps_font::ps_font,
	subencoding::subencoding, ps_printer::define_encoding,
	ps_printer::encode_fonts):
	* src/devices/grops/psrm.cpp (resource_manager::document_setup,
	resource_manager::supply_resource):
	* src/devices/grotty/tty.cpp (tty_printer::tty_color,
	tty_printer::tty_printer, tty_printer::color_to_idx,
	tty_printer::add_char):
	* src/include/itable.h (ITABLE(T)::~ITABLE(T),
	ITABLE(T)::define):
	* src/include/ptable.h (PTABLE(T)::~PTABLE(T),
	PTABLE(T)::define):
	* src/libs/libbib/index.cpp
	(index_search_item::~index_search_item, make_index_search_item,
	index_search_item_iterator::index_search_item_iterator,
	index_search_item::get_tag, index_search_item::munge_filename,
	index_search_item::search):
	* src/libs/libbib/linear.cpp (bmpattern::~bmpattern,
	file_buffer::file_buffer, file_buffer::load,
	linear_searcher::linear_searcher,
	linear_searcher::~linear_searcher):
	* src/libs/libbib/search.cpp
	(search_list_iterator::search_list_iterator,
	search_item::search_item):
	* src/libs/libdriver/input.cpp (IntArray::~IntArray,
	IntArray::append, StringBuf::~StringBuf, StringBuf::append,
	get_integer_arg, parse_x_command, do_file):
	* src/libs/libdriver/printer.cpp (printer::printer,
	printer::load_font, text_file::~text_file, text_file::next,
	font::~font, font_widths_cache::~font_widths_cache,
	font:alloc_ch_index, font::extend_ch, font::compact,
	font::load_desc):
	* src/libs/libgroff/fontfile.cpp (font::open_file):
	* src/libs/libgroff/relocate.cpp (searchpath, searchpathext,
	set_current_prefix):
	* src/libs/libgroff/searchpath.cpp (search_path::search_path,
	search_path::command_line_dir, search_path::open_file,
	search_path::open_file_cautious):
	* src/libs/libgroff/string.cpp (sfree, srealloc,
	string::remove_spaces):
	* src/libs/libgroff/symbol.cpp (symbol::symbol, concat):
	* src/libs/libgroff/tmpfile.cpp (temp_init::temp_init,
	xtmpfile_list_init::~xtmpfile_list_init, xtmpfile):
	* src/preproc/eqn/box.cpp (set_gfont, set_grfont, set_gbfont,
	box_list::append, box_list::~box_list):
	* src/preproc/eqn/delim.cpp (make_delim_box,
	delim_box::~delim_box):
	* src/preproc/eqn/eqn.ypp (number):
	* src/preproc/eqn/lex.ypp (file_input::~file_input,
	argument_macro_input::~argument_macro_input):
	* src/preproc/eqn/pile.ypp (matrix_box::~matrix_box,
	matrix_box::append):
	* src/preproc/eqn/special.cpp (special_box::~special_box):
	* src/preproc/eqn/text.ypp (set_char_type):
	* src/preproc/html/pre-html.cpp (get_line, scanArguments):
	* src/preproc/pic/object.cpp (output::~output,
	output::set_args, text_item::~text_item,
	object_spec::~object_spec, command_object::~command_object,
	line_object::~line_object):
	* src/preproc/pic/pic.ypp (placeless_element, reset_variables,
	print_args, text_expr, object_spec, text, sprintf_args, path):
	* src/preproc/refer/command.cpp (input_item::~input_item,
	input_item::peek_char):
	* src/preproc/refer/label.ypp (lookup_label):
	* src/preproc/refer/refer.cpp (store_citation, store_reference):
	* src/preproc/tbl/main.cpp (format::add_rows, format::~format):
	* src/preproc/tbl/table.cpp (block_entry::~block_entry,
	table::~table, table::allocate):
	* src/roff/groff/groff.cpp (possible_command::~possible_command,
	possible_command::clear_name):
	* src/roff/troff/column.cpp
	(justification_spec::~justification_spec,
	justification_spec::append):
	* src/roff/troff/dictionary.cpp (dictionary::lookup):
	* src/roff/troff/env.cpp (override_sizes, tab_stops::to_string,
	hyphen_word, hyphen_trie::insert_hyphenation):
	* src/roff/troff/input.cpp (read_long_escape_name, token::next,
	do_get_long_name, temp_iterator::~temp_iterator, get_delim_name,
	pipe_source, read_string, pipe_output, system_request,
	open_mac_file, do_macro_source, do_register_assignment,
	do_string_assignment, read_draw_node, copy_mode_error):
	* src/roff/troff/node.cpp (troff_output_file::set_font,
	troff_output_file::~troff_output_file, draw_node::is_tag,
	grow_font_table, font_family::~font_family,
	font_family::make_definite):
	* src/utils/hpftodit/hpftodit.cpp (name_list::~name_list,
	read_map):
	* src/utils/indxbib/indxbib.cpp (main, get_cwd):
	* src/utils/tfmtodit/tfmtodit.cpp (tfm::~tfm, tfm::load): Port
	uses of `a_delete` to `delete[]`.

2021-08-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Drop `ad_delete` preprocessor wrapper for the
	`delete` operator.  It has been rendered unnecessary by the
	removal of support for ancient C++ compilers.

	* src/include/lib.h (ad_delete): Drop symbol definition.

	* src/preproc/pic/object.cpp (graphic_object::print_text):
	* src/preproc/refer/ref.cpp (reference::reference,
	reference::merge, reference::insert_field,
	reference::delete_field):
	* src/preproc/tbl/main.cpp (format::add_rows): Port uses of
	`ad_delete` to `delete[]`.

2021-08-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	* m4/groff.m4 (GROFF_ARRAY_DELETE): Update `AC_MSG_CHECKING`
	argument to refer to "ISO C++98" instead of "ANSI".  Call
	`AC_MSG_ERROR` (aborting configuration) if the check fails.

2021-08-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Drop support for `ARRAY_DELETE_NEEDS_SIZE`.

	* src/include/lib.h [ARRAY_DELETE_NEEDS_SIZE]: Drop preprocessor
	conditional branch.  This abandons support for certain pre-ISO
	C++98 compilers.  (According to a now-removed comment, unsized
	array deletion was documented in "ARM", meaning _The C++
	Annotated Reference Manual_, published in 1989.)

	* Makefile.am: Undocument preprocessor symbol.

2021-08-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Drop redefinition of `INT_MIN`.

	* src/include/lib.h: Drop redefinition of `INT_MIN`.  It was
	motivated by a bug in the AT&T C++ compiler (cfront), version
	2.0, released in June 1989.  Implementations have had 30 years
	to get this right; assume that they have.

	* PROBLEMS: Drop corresponding item.

2021-08-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Drop long-dead `COOKIE_BUG` logic.

	* src/libs/libgroff/new.cpp (operator new, operator delete):
	Drop preprocessor conditional branches on the `COOKIE_BUG`
	symbol.  This was de-documented, and an autoconf check for its
	necessity removed, way back in groff 1.10 (November 1995).  It
	was relevant only to GCC from versions 2.0 to 2.2.2 (all 1992).

2021-08-24  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Refactor `font::scan_papersize()`.

	* src/libs/libgroff/font.cpp (font::scan_papersize): Ensure
	successful return value of `fgets()` before attempting to parse
	string for paper format.  Convert `test_file` to Boolean and
	rename to `attempt_file_open`.  Quiets GCC `-Wunused-result`
	warning.

2021-08-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	[tests]: Apply naming convention.

	* tmac/tests/e_footnotes_work_with_columns.sh: Rename from...
	* tmac/tests/e_footnotes-work-with-columns.sh: ...to this.
	* tmac/tmac.am (tmac_TESTS): Update.

2021-08-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an.tmac (IP): Make code more accessible by using a
	Boolean operator rather than an arithmetic one (assembly
	language programmers are comfortable with the interchangeability
	of subtraction and comparison, but not everyone is).

2021-08-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	groff(7): Document more escapes.

	* man/groff.7.man (Escape short reference): Document
	`\[charNNN]`, `\[uNNNN]`, `\[uNNNN_NNNN...]` escape sequences.
	Revise descriptions of `\O0` and `\O1` escape sequences.
	Document `\O2`, `\O3`, `\O4`, and `\O5` escape sequences.

2021-08-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	groff(7): Fix bug in `\?` escape rendering.

	* man/groff.7.man (ESC?): Fix bug in page-private macro: the
	question mark that ends this uniquely-syntaxed escape sequence
	was not shown with the mandatory escape character before it.
	Problem dates back at least to commit 41b0e794, 19 February
	2007, but possiby to f790bc7a9, 6 January 2002.

2021-08-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	[docs]: Elaborate macro definition discussion.

	Fixes <https://savannah.gnu.org/bugs/?57944>.  Thanks to Dave
	Kemper for the report.

2021-08-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	[tests]: Make robust against $GROFF_TYPESETTER settings.

	* src/roff/groff/tests/\
	use_point_size_escape_with_single_digit_arg.sh:
	* src/roff/nroff/tests/verbose_option_works.sh:
	* tmac/tests/s_IP-indents-using-paragraph-type-size.sh: Export
	an empty $GROFF_TYPESETTER to the environment.

	Fixes <https://savannah.gnu.org/bugs/?61040>.  Thanks to Bjarni
	Ingi Gislason for the report.

2021-08-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Treat an empty $GROFF_TYPESETTER as unset.

	* src/libs/libgroff/device.cpp (device_init::device_init): Test
	both returned pointer from `getenv()` and, if that's not null,
	the first character of the string for nullity before assigning
	it to `device`.

	Fixes <https://savannah.gnu.org/bugs/?61068>.

2021-08-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	Update .version file more aggressively.

	* Makefile.am: Regenerate a temporary version string file on
	every build, but update the real version file's mtime only if
	its contents change.
	(.version.tree): New target.  Mark it phony, populating it with
	`$(VERSION)` with every `make` if the `.tarball-version` file
	{created by `dist-hook`} doesn't exist.
	($(top_srcdir)/.version): Depend on foregoing target.  If the
	tarball version file and the target do not exist, copy `.tree`
	file into place.  Compare this file with `.tree` file; if they
	differ, move the `.tree` file to this one.
	(MOSTLYCLEANFILES): Add `.version.tree`.

	Fixes <https://savannah.gnu.org/bugs/?61052>.  Thanks to Bjarni
	Ingi Gislason for the report.

2021-08-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff]: Rename font class member variable.

	* src/include/font.h (font class): Rename `unscaled_charwidths`
	member variable to `use_unscaled_charwidths`: since it is a
	Boolean, make it read more like a logical predicate.

	* src/libs/libgroff/font.cpp (font::get_width, font::load_desc):
	* src/libs/libgroff/fontfile.cpp (font class): Update
	initialization and assignments.

2021-08-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (get_copy): Demote integer arguments
	to Booleans and use Boolean literals for default values.  Rename
	`defining` to `is_defining`.
	(get_char_for_escape_name, do_define_macro): Update call sites
	that use non-default arguments to use Boolean literals and
	comments to document what's being requested.

2021-08-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/doc.am (DOC_GROFF_ONLY): Stop running groff in unsafe
	mode; no documents (in _this_ directory) appear to require it.

2021-08-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grn]: Fix code style issues.

	* src/preproc/grn/hgraph.cpp:
	* src/preproc/grn/hpoint.cpp:
	* src/preproc/grn/main.cpp:
	* src/preproc/grn/hdb.cpp: Drop use of `register` storage class.

	* src/preproc/grn/hgraph.cpp (len, HGPrintElt, picurve):
	* src/preproc/grn/hdb.cpp (DBRead): Wrap long lines.

	* src/preproc/grn/hgraph.cpp: Rename function from
	`Paramaterize` to `Parameterize`.
	(HGCurve): Update call site.

	* src/preproc/grn/main.cpp (add_file): Drop redundant cast in
	`realloc()` call.
	(conv, interpret): Use standard English in diagnostic messages.

2021-08-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/grn/hdb.cpp: Perform more input validation.
	Improve diagnostics by taking advantage of libgroff
	infrastructure and tracking the line number of the input file.
	Add global `lineno`.
	(DBRead): Increment `lineno` after reading newlines from input.
	Call `error_with_file_and_line()` instead of `error()`.  If
	input reports a negative length for the text (string) to follow
	in the file, exit with a fatal diagnostic.  Check for EOF while
	reading text string.
	(DBGetType): Convert `fprintf()` call for warning diagnostic to
	`warning_with_file_and_line()`.
	(DBRead, DBGetType): Add contextual information to diagnostic
	messages.

	Fixes <https://savannah.gnu.org/bugs/?61043>.  Thanks to
	Savannah user eqkws for the report.

2021-08-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/grn/main.cpp (usage): Update usage message.

2021-08-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grn]: Add and use `malloc()` wrapper.

	* /src/preproc/grn/main.cpp (grnmalloc): New function takes
	argument of `size_t` type and constant string argument to
	describe what is being allocated.  Return non-null pointer from
	`malloc()`, otherwise call `fatal()`, describing what was being
	allocated and the problem reported by the system.

	* src/preproc/grn/hdb.cpp (DBCreateElt):
	* src/preproc/grn/hpoint.cpp (PTMakePoint):
	* /src/preproc/grn/main.cpp (main, interpret): Migrate
	`malloc()` callers to `grnmalloc()`.

2021-08-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/grn/hdb.cpp (DBRead): Check return value of
	`sscanf()` and call `fatal()` if no conversions succeeded.  The
	blithe discard of a useful return value is bad enough, but this
	one took place inside a do-while such that it could loop
	forever trying fruitlessly to parse two doubles out of strings
	that didn't contain them (the loop never checked the EOF status
	of the file stream from which it was reading, and relied on
	`fgets()` to keep advancing the stream pointer).  Discovered
	while root-causing Savannah #61043.

2021-08-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	Resolve compiler warnings relating to format string security and
	ISO C++98 conformance.

	* src/preproc/eqn/delim.cpp (define_extensible_string):
	* src/preproc/pic/pic.ypp (do_sprintf): Use #pragma to silence
	GCC "format-nonliteral" warning and explain why our usage is
	safe in a comment.

	* src/preproc/preconv/preconv.cpp (detect_file_encoding): Use
	`l` modifier to `%u` `fprintf()` conversion instead of `z`, and
	cast return values of `size_t` to unsigned long; "ISO C++98
	does not support the ‘z’ gnu_printf length modifier" (it doesn't
	support `ll` either).  N.B. this is debugging output only.

	* src/roff/groff/groff.cpp (synopsis):
	* src/roff/troff/input.cpp (usage): Repeat variadic argument for
	`%s` conversion in `fprintf()` call because "ISO C++98 does not
	support %n$ operand number formats".

	* src/roff/troff/env.cpp: Initialize adjustment and hyphenation
	mode enums without commas at the end of the enumeration lists.

2021-08-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Clean up the interface between `an-footer` and `BT`, and
	HTML header and footer suppression generally.  Move
	responsibility for checking the `ps4html` register from the
	latter to the former.  This is not something a `BT` redefiner
	{see groff_man(7)} should have to worry about.

	* tmac/an.tmac (BT): Drop test of `ps4html` register and early
	return.
	(an-header, an-footer): Return immediately if
	`an-suppress-header-and-footer` is true.  Remove conditional on
	`an-is-output-html`.
	(initialization): Define `an-suppress-header-and-footer`
	register; true if `an-is-output-html` is true, or if `ps4html`
	is defined.

2021-08-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/html/pre-html.cpp (get_resolution): Unbrace a
	single-statement `while` loop per prevailing coding style.  Add
	comments noting how we aren't parsing DESC files _precisely_ as
	our documentation specifies.

2021-08-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	Convert font class `int` members to `bool` where appropriate.

	* src/include/font.h (font class): Demote integers to Booleans
	and use Boolean literals where possible.  Rename some member
	variables to look like logical predicates.
	  - contains()
	  - is_special()
	  - has_ligature()
	  - tcommand -> has_tcommand
	  - unscaled_charwidths
	  - pass_filenames
	  - use_charnames_in_special
	  - is_unicode
	  - special (private)
	* src/libs/libgroff/font.cpp (font::font): Construct object
	using Boolean literal.
	(font::contains):
	(font::is_special):
	(font::has_ligature): Convert return type to `bool`.
	(font::load):
	(font::load_desc): Assign to member variables using Boolean
	literals.
	* src/libs/libgroff/fontfile.cpp: Initialize appropriate globals
	using `bool` type and Boolean literals.
	* src/roff/troff/troff.h:
	* src/roff/troff/input.cpp (main):
	* src/roff/troff/node.cpp (troff_output_file::put_char_width):
	Rename `tcommand_flag` to `device_has_tcommand` to suggest a
	logical predicate, and to emphasize the formatter's concern with
	what the output device can accept in the intermediate output
	language.
	* src/roff/troff/input.cpp: Initialize `device_has_tcommand`
	global using `bool` type and Boolean literal.

2021-08-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	[me]: Clarify breadth of application of `ll` macro.

	* doc/meref.me: Document application of .ll macro only to the 3
	environments that me(7) uses, not any the user might create.
	* tmac/e.tmac (@C, xl, ll): Update comments.

	Fixes <https://savannah.gnu.org/bugs/?55060>.

2021-08-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/e.tmac (1c): Fix problem when combining multiple columns,
	footnotes, and return to single-column mode: do not place the
	single-column text below the footnote.  Space to the end of the
	page (\n[nl]), not the page "high-water mark" (\n[.h]).  Thanks
	to Bjarni Ingi Gislason for the suggestion.

	Fixes <https://savannah.gnu.org/bugs/?58736>.

2021-08-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #58736.

	* tmac/tests/e_footnotes_work_with_columns.sh: Test it.
	* tmac/tmac.am (tmac_TESTS): Run test.

2021-08-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/e.tmac (bc): Fix problem with multiple columns on long
	pages.  Space to the next page location trap, not 24 inches.
	Thanks to Dave Kemper and Bjarni Ingi Gislason for help tracking
	this bug down.

	Fixes <https://savannah.gnu.org/bugs/?55081>.

2021-08-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #55081.

	* tmac/tests/e_columns-work-on-long-pages.sh: Test it.
	* tmac/tmac.am (tmac_TESTS): Run test.

2021-08-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/e.tmac: Validate page length to prevent too-short paper
	formats from causing traps to call each other until the input
	stack overflows.  The minimum is 14v.
	(check_page_length): Define new macro to perform the check.
	(bp, @p, sh, 2c, 1c): Call it from likely candidates (@p is
	internal and called by paragraphing macros).
	(pl): Intercept and wrap the `pl` request itself; check its
	setting immediately.

	Fixes <https://savannah.gnu.org/bugs/?61034>.

2021-08-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff.7.man (Localization): Add section.

2021-08-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/cs.tmac:
	* tmac/de.tmac:
	* tmac/fr.tmac:
	* tmac/it.tmac:
	* tmac/sv.tmac: Update ms package localization to use an
	appropriate hyphenation mode for the `HY` register (the
	mode for the pattern files, plus 2 as is traditional in ms).

2021-08-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/e.tmac (1c): Place comment escape sequence immediately
	after macro argument.  (When redefining a request, one needs to
	remember that request arguments are not parsed exactly as macro
	arguments are.)  Quiets "-w tab" warning.

2021-08-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	[me]: Fix misleading documentation claim; gremlin(1) pictures
	cannot be included directly; they must be preprocessed with
	grn(1).

	* doc/meref.me:
	* tmac/groff_me.7.man: Replace "gremlin" with "grn".

2021-08-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_me.7.man: Document `n1` and `n2` macros.

2021-08-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	[me]: Fix PS/n1 combination diagnostic spew and apparent logic
	error.

	* tmac/e.tmac (@h): Place comment escape sequence immediately
	after macro argument.  (When redefining a request, one needs to
	remember that request arguments are not parsed exactly as macro
	arguments are.)
	(n2): Prefix with `do` requests using long register names.
	(n2): Add macro-local register `|l` to store length of first
	argument.  Don't attempt to extract a substring beyond its
	bounds, which produces a warning diagnostic.
	(&&): Define no-op macro for use as end macro in nested macro
	definition.
	(PS): Fix unbalanced-else diagnostic by changing `if` to `ie`.
	(PS): Initialize `PS_nm_cnt` register to 0 instead of relying on
	implicit definition.
	(PS): Prefix with `do` definition of long macro name.
	(PS): Indent call of end macro.
	(PS): Add comment after escaped space.
	(PS, PF): Prefix with `do` call of long macro name.

	Fixes <https://savannah.gnu.org/bugs/?61019>.

2021-08-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.pl (do_line): Recognize `n1` and `n2` as
	characteristic me(7) macros.

2021-08-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_me.7.man: Document GF, IE, IF, IS, and PF macros.

2021-08-08  Dave Kemper <saint.snit@gmail.com>

	* tmac/e.tmac: Rename `PE` to `PF` but remove vertical spacing.
	(PE): New `PE` calls `PF` and then space as old `PE` did.

	Fixes <https://savannah.gnu.org/bugs/?60479>.

2021-08-06  Dave Kemper <saint.snit@gmail.com>

	* doc/meref.me: Improve documentation of image-inclusion macros.

	Fixes <https://savannah.gnu.org/bugs/?60478>.

2021-08-06  Dave Kemper <saint.snit@gmail.com>

	GNU pic recognizes two possible endings of a pic block: .PE or
	.PF.  This fact was documented in doc/pic.ms but not in the pic
	man page.  The minimal pic implementation provided by pic.tmac
	also did not include a .PF definition.

	* src/preproc/pic/main.cpp (main): Add dummy definition of `PF`
	macro.
	* tmac/pic.tmac (PF): Add macro that performs indentation only.
	(PE): Call PF for indentation.

	* doc/pic.ms (Interface to [gt]roff/How Scaling is Handled):
	* src/preproc/pic/pic.1.man (Description, Options): Mention `PF`
	in addition to `PS` and `PE`.

	Fixes <https://savannah.gnu.org/bugs/?60504>.

2021-08-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/refer-ms.tmac:
	* tmac/s.tmac (fn@init, @PP): Rename strings
	`fn@sup-{start,end}` to `fn@mark-{start,end}.  This makes it
	more clear that the strings are brackets for the footnote
	marker, and not so much to do with the note text.  Further, in
	nroff mode, they aren't superscripts at all, but bracket glyphs.

2021-08-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/refer/refer.1.man (Examples): Add section.

	Addresses part of <https://savannah.gnu.org/bugs/?57667>.

2021-08-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/refer-ms.tmac: Use `fn@sup-{start,end}` instead of
	`par@sup-{start,end} when setting refer(1) citation labels.
	This improves the appearance of the default labels on nroff
	devices.  Continues commit caeede07, 1 May.

2021-08-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tman/an-ext.tmac (EX, EE): Stop manipulating hyphenation.
	It's redundant since disabling fill mode already prevents
	automatic breaking, and therefore automatic hyphenation.
	* tmac/groff_man.7.man.in (Description/Document structure
	macros) <EX, EE>: Update documentation.

2021-08-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.pl: Update authorship credits.  I appear
	to have removed or refactored all of the "device" (recte: macro
	package) inference logic adapted from Ralph Corderoy's grog.sh;
	see commit 3617f42048f54cc3f0adc282ee3b9e481c75ebd5, 13 June
	2014.

2021-07-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.pl: Refactor through relocation and
	renaming.  Move several global objects into subroutines.
	  - `@request` -> `&do_line`
	  - `@macro_ms`, `@macro_man`, `@macro_man_or_ms` ->
	    `&infer_man_or_ms_package`
	  - `@main_package` -> `&construct_command`
	Rename some objects for clarity.
	  - `%Groff` -> `%score`
	  - `@filespec` -> `$input_file`
	Delete unused object.
	  - `@standard_macro`
	Add comments.

2021-07-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.pl (do_line): When matching macro/request
	names, accept any non-whitespace character (Perl: \S) instead of
	just a word-constituent character (Perl: \w), since roff
	identifiers can contain unusual characters (refer(1) and
	groff_me(7) illustrate some real-world cases).

2021-07-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.pl: Refactor macro package inference.
	  - Stop manually populating `Groff` hash.  (It's itching for a
	    rename.)
	  - Stop calling now-dead subroutine `infer_macro_packages`.
	(do_line): Always store the names of all macros called to the
	`Groff` hash, incrementing it.  Now it's a proper scoreboard.
	Populate `inferred_main_package` from here upon encountering
	characteristic (i.e., uniquely named) macros of each package.
	Bug fix: drop "SP" from list of characteristic mom(7) macros;
	it's shared with mm(7).
	(infer_man_or_ms_package): Update comment.  Return 0 (false) if
	document appears to be "raw", using no full-service macro
	package.
	(infer_macro_packages): Delete; do_line() does this work.

2021-07-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.pl: Refactor preprocessor inference.
	  - Add new list, `inferred_preprocessor`.
	  - Drop preprocessor-related keys from `Groff` hash.
	  - Drop scalar `inside_tbl_table`.
	(do_line): Set up hash `preprocessor_for_macro`.  Detect
	preprocessor macros the way the preprocessors do, explained in a
	comment.  Respect AT&T compatibility mode when doing so.  Build
	list of inferred preprocessors.  This replaces the extensive and
	gaseous series of `if` statements that manipulated the `Groff`
	hash.
	(infer_preprocessors): Completely replace.  Set up a hash
	`option_for_preprocessor` mapping preprocessor names to groff
	options (where applicable).  Append to `command` and
	`preprocessor` lists as appropriate.  Sort the preprocessor
	options so they don't move around in the argument list depending
	on the order of their macros' appearance in the input.

	* src/utils/grog/tests/smoke-test.sh: Update test cases to
	expect preprocessor options to show up in sorted order.

2021-07-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.pl: Drop dead code.  Delete global
	hash `preprocs_tmacs`, unused since commit b0de53c9, 30 June.

2021-07-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grohtml]: Fix code nit.

	* src/preproc/html/pre-html.cpp (get_image_generator):
	Type-qualify `keyword_len` as `const` to encourage the compiler
	to compute this value (the length of a string literal) at build
	time.

2021-07-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Slightly refactor.

	* src/roff/troff/node.cpp: Add static `image_filename_len` to
	store length of cached image file name, so we don't wastefully
	recompute its length every time one is re-used.
	(suppress_node::tprint): Compute image file name length only
	when saving a new image file name.  Use saved length in later
	expressions.
	(min): Drop unused inline function.

2021-07-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Fix regression (breaking important pdfmark/pdfroff
	features) caused by my commit
	e876d4bfd193abb9a7d1fb6e76519349bded482a, 27 July.  An empty
	image file name is in fact semantically valid when writing a
	bounding box; pdfmark uses such boxes to frame hyperlinks.
	Avoid null pointer dereference by initializing `image_filename`
	to an empty string literal.  Thanks to Tadziu Hoffman and Keith
	Marshall for their patient explanations.

2021-07-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Refactor some internal functions.

	* src/roff/troff/node.cpp: Rename two functions that have only
	one caller.
	  - get_reg_int -> get_register
	  - get_reg_str -> get_string
	We now have:
	(get_register, get_string): Use `assert()` aggressively because
	these are deeply internal and validation is imperative.  Get rid
	of diagnostic messages (one of which perpetuated the dubious
	"number register" nomenclature) accordingly.
	(fetch_register): Also rename `prev_value`, misleadingly
	specific and seemingly copy-and-pasted out of a context where an
	auto-increment or -decrement might have been applied.  It's just
	the `value`.
	(suppress_node::tprint): Update call site.

2021-07-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Refactor `get_value` member functions.  Given their
	names and popular "getter/setter" paradigms from many OO
	languages, the return type is misleading.  Change it from `int`
	to `bool` since it returns only a success/failure status and
	modifies an argument (passed by reference) to deliver the
	requested data.

	* src/roff/troff/reg.h (reg, variable_reg):
	* src/roff/troff/div.cpp (page_offset_reg, page_length_reg,
	vertical_position_reg, high_water_mark_reg,
	distance_to_next_trap_reg, page_number_reg, no_space_mode_reg):
	* src/roff/troff/env.cpp (int_env_reg, vunits_env_reg,
	hunits_env_reg, horizontal_place_reg):
	* src/roff/troff/input.cpp (writable_lineno_reg):
	* src/roff/troff/reg.cpp (reg, number_reg, variable_reg): Update
	class and member function definitions to reflect the new type.
	Update member function definitions to return appropriate Boolean
	literals instead of 0 and 1.

2021-07-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grohtml]: Fix Savannah #60981.  Refactor device description
	file handling and make it more robust.  Make pre-grohtml's
	-F option work as documented and honor device description file
	semantics documented in groff_font(5).

	* src/preproc/html/pre-html.cpp: Store partial filespecs of HTML
	and PostScript device files in constant strings.
	(get_resolution): Initialize `res` to zero.  Only `free()` the
	pointer `pathp` if `open_file()` succeeded (populating it).
	Stop checking for valid `sscanf()` conversions of an integer to
	store in `res`, and don't return early.  Instead process the
	whole DESC file; per our Texinfo manual and groff_font(5),
	"Later entries in the file ... override previous values."
	(get_image_generator): Add new function, paralleling
	`get_resolution()`, replacing open-coded logic in `main()`.
	Only the body of the `while` loop significantly differs.
	Instead of using `sscanf`, process the input character by
	character matching the keyword and {1,} spaces or tabs
	after it.  Whatever follows in `linebuf` must be the program
	name.
	(imageList::createPage): Add `assert()` to cause failure if an
	empty image generator program gets this far, because it creates
	repeated messes if it does.  (Something isn't checking exit
	statuses.)
	(main): Condense `image_gen` collection to a function call, a
	null pointer check, and a fatal diagnostic.  Add a sanity check
	and a fatal diagnostic if the PostScript resolution is garbage.
	{It's initialized to -1 and `get_resolution()` will return 0 if
	it doesn't find one.}

	Fixes <https://savannah.gnu.org/bugs/?60981>.

2021-07-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grohtml]: Delete or rename `debug` symbols to avoid clash with
	new `debug` function in libgroff.

	* src/preproc/html/pre-html.cpp: Update a comment to refer to
	"debugging" mode.  Rename static global from `debug` to
	`debugging`.
	(html_system, imageList::createPage, imageList::createImage,
	print_args, char_buffer::do_html, char_buffer::do_image,
	scanArguments): Update uses of static global.

	* src/preproc/html/pushback.cpp:
	* src/preproc/html/pushback.h: Delete unused `debug` member
	variable.

2021-07-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/nroff/nroff.sh: Fix straggling use of `$1`, which
	might not have a clear meaning in some shells after being
	shifted (what if we just shifted the last argument?).  Also
	`continue` after processing a (groff) option argument, as we
	should make no attempt to parse it.

2021-07-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (do_suppress): Recast diagnostic
	messages for clarity, to communicate in terms of what the
	escape sequence _is_, and to quote literals expected as input.
	* src/roff/troff/node.cpp (suppress_node::tprint): Recast
	diagnostic to prevent misinterpretation of image "description"
	as being HTML IMG tag alt text or something like that.
	Unfortunately, the new language speaks in terms of internal
	implementation details ("grohtml-info"), but at least that way
	it's easier for a user to decide if they care about it.

2021-07-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Avoid using sprintf() with user-controlled format
	string.

	* src/preproc/html/pre-html.cpp (makeFileName): Add comment
	noting need for implementation synchrony between this function,
	which generates \O5 escape sequences, and troff's
	suppress_node::tprint member function, which interprets them.
	* src/roff/troff/node.cpp: Rename 2 static globals for clarity.
	  - `last_image_filename` -> `image_filename`
	  - `last_image_id` -> `subimage_counter`
	(suppress_node::tprint): Set up the buffer for image file name
	to be written using a constant rather than an embedded literal.
	Unconditionally initialize the buffer with a string terminator,
	so there is no chance of a read from unitialized storage.  Drop
	unused code involving `tem`.  Drop stale comments.  Clarify
	comment: an `image_filename` doesn't _always_ contain a format
	string, only sometimes.  Replace use of `sprintf` with manual
	construction of a new image filename string.  There are two
	cases, one where a format string {presently "%d"} is present,
	and one where it is not.  If it is present, locate it
	{`percent`}.  This means a limited/bounded image ("subimage") is
	being processed; increment the subimage counter.  Write a new
	image file name preserving the parts before and after "%d" (the
	"prefix" and "suffix", and replacing only the middle, using
	`sprintf` with the subimage counter and the (string literal)
	format.  Be mindful of string bounds and memory allocation,
	issuing diagnostics or aborting as necessary.  If the image file
	name does _not_ contain a format string, but needs only to be
	copied, do that (`strcpy`), again instead of using `sprintf`.

	Fixes <https://savannah.gnu.org/bugs/?60977>.

2021-07-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grohtml]: Fix Savannah #60971.

	* src/preproc/html/pre-html.cpp (makeFileName): Consistently put
	a dash at the end of `macroset_template` whether the image file
	name stem is user-supplied or the default.  Stop adding the dash
	before the image number in `image_template` instead.  This makes
	the image file name format reliable whether the image needs to
	be subdivided (eqn) or not (tbl).

	Fixes <https://savannah.gnu.org/bugs/?60971>.

2021-07-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grohtml]: Reduce noise to standard error stream.

	* src/preproc/html/pre-html.cpp (imageList::createImage): Fix
	apparent oversight: `EXE_EXT` was not being applied to the
	second of three commands being run in a pipeline (did Windows
	users notice?).  Also issue (undocumented!) `-quiet` option to
	`pnmtopng` to shut up its commentary about counting colors.

2021-07-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grohtml]: Stop discarding standard error output.

	* src/preproc/html/pre-html.cpp (html_system): Stop sending the
	standard error stream to /dev/null; this just makes debugging
	harder, and doesn't produce much more output for well-formed
	input documents.  Rename `save_stdout` to `saved_stdout`.

2021-07-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grohtml]: Fix misspelled variable.

	* src/preproc/html/pre-html.cpp: Rename `IMAGE_BOARDER_PIXELS`
	to `IMAGE_BORDER_PIXELS`.

2021-07-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	[libgroff, troff]: Add debug diagnostic level.

	* src/include/error.h: Declare functions
	`debug_with_file_and_line` and `debug`.
	* src/libs/libgroff/error.cpp: Add `DEBUG` to enum `error_type`.
	(do_error_with_file_and_line): Add case for `DEBUG` in switch.
	(debug, debug_with_file_and_line): Add new functions.

	Do the same for troff since it has a private implementation of
	the diagnostic functions (thanks to `output_warning()`).

	* src/roff/troff/input.cpp: Add `DEBUG` to enum `error_type`.
	(do_error_with_file_and_line): Add case for `DEBUG` in switch.
	(debug, debug_with_file_and_line): Add new functions.

2021-07-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/hpftodit/hpftodit.cpp (hp_msl_to_ucode_name,
	unicode_to_ucode_name): Hush "format nonliteral" compiler
	warnings by using a preprocessor-defined string literal as an
	sprintf() format string instead of a C++ variable that just
	compared identically to the same thing.

2021-07-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/hpftodit/hpftodit.cpp (show_symset): Prevent
	sprintf() from overunning a static buffer.  Thanks to Bjarni
	Ingi Gislason for the report.  Resize buffer to be large enough
	to accommodate a 64-bit unsigned int type formatted as decimal.
	Also add assert() before the sprintf() to abort if the int type
	is even larger than that.  Use "%u" conversion instead of "%d"
	since the quantity is unsigned.
	(hp_msl_to_ucode_name): Similar, but for a signed int.
	(unicode_to_ucode_name): Similar, but for a signed int formatted
	as (unsigned) hexadecimal.

	Fixes <https://savannah.gnu.org/bugs/?60964>.

2021-07-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/nroff/nroff.sh: Refactor, mostly by relocation.  Move
	locale character set inference logic so that we only run it if
	we have to (`-T` option was invalid or not given and
	`GROFF_TYPESETTER` was not set in the environment or invalid).
	Initalize `T` variable as empty.

2021-07-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/nroff/nroff.sh: Use Version 7 Unix sh-compatible form
	of parameter expansion; per the GNU Autoconf manual[1], "[o]ld
	BSD shells, including the Ultrix sh, don't accept the colon for
	any shell substitution, and complain and die."  This is also the
	form of such substitutions used elsewhere in the script, so they
	should be mutually consistent.

	[1] https://www.gnu.org/software/autoconf/manual/autoconf-2.60/\
	html_node/Shell-Substitutions.html

2021-07-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/nroff/nroff.sh: Slightly refactor.  Rename formerly
	unused loop index variable from `i` to `arg` to leverage it in a
	later diagnostic message without a separate definition.  Inside
	the loop, expand `arg` instead of positional parameter 1, the
	meaning of which is less obvious after the parameter list is
	shifted by the new `is_option_argument_pending` logic (commit
	2b955c57, 20 July).

2021-07-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/groff.1.man (Options) <-S>: Fix error: `-S`
	sets safer mode (redundantly) but is not passed to pic and troff
	as was claimed.

2021-07-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/number.cpp (parse_term): Tweak diagnostic
	messages.  When a left operand to a binary operator is empty,
	report the operator in question (helpful for string
	interpolations in complex expressions).  When handling input
	scaling indicators, drop word "this" from "this context", since
	the parser's context might not be clear from a file name and
	line number.  Say "scaling indicator" instead of "scale
	indicator".
	(parse_expr, parse_term): Rename function parameter from
	`scale_indicator` to `scaling_indicator`.

2021-07-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (define_color): Tweak warning
	diagnostics for clarity.  Only one value is expected when
	defining a color in the `gray` color space.  Quote recognized
	color space names when complaining of an unknown one.

2021-07-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	[nroff]: Pass a supplied `-c` option to groff.  While at present
	this does little more than `-P-c` already achieves, that might
	not be true forever.  Further, there is a distinction between
	ignoring color-related requests in groff input documents and
	ignoring color-related commands in device-independent output.
	This distinction would make it unwise to adapt `-c` to any other
	purpose in nroff in the future.

	* NEWS: Update item from commit dd725dce, 10 February 2020.
	* src/roff/nroff/nroff.sh: Pass `-c` option to groff in addition
	to synthesizing a `-P-c` option (for grotty) when it is seen.
	* src/roff/nroff/nroff.1.man (Options): Document `-c` as an
	option recognized by troff(1).

2021-07-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	[nroff]: Support space between option flags and their arguments.

	* src/roff/nroff/nroff.sh: Add new `is_option_argument_pending`
	flag to add an additional state to the option parser.  Set it if
	an argument-requiring option is encountered with no abutting
	argument.  If it is set when a new argument is encountered,
	absorb the argument into the groff option list and clear the
	flag.  Modernize script in a couple of other ways.
	  - Exit with status 2 upon usage errors so that this condition
	    is easily distinguished from a groff abort (which exits with
	    status 1).
	  - Test shell variables for non-nullity with test(1) -n
	    operator instead of a string comparison.
	* src/roff/nroff/tests/verbose_option_works.sh: Test correct
	construction of groff command using this input form.
	* src/roff/nroff/nroff.1.man (Options): Add spaces between
	option flags and option arguments.  Sync metasyntactic variable
	names with groff(1) while we're at it.
	(Description): Drop sentence warning of whitespace prohibition.
	(Exit status): Add new section.

	Fixes <https://savannah.gnu.org/bugs/?60943>.

2021-07-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (main): Emit error diagnostic if `-d`
	or `-r` option is given a malformed argument.  This prevents a
	string or register with an empty name from being created.

	Fixes <https://savannah.gnu.org/bugs/?60935>.

2021-07-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/nroff/nroff.sh:
	* src/roff/troff/input.cpp (usage): Use "dev" as metasyntactic
	variable name for -T option argument to achieve consistency with
	our other documentation.

2021-07-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	[groff]: Comment out vestiges of gideal support.  It would be
	lovely to have, but no free (and GPL-compatible) implementation
	has showed up after decades of waiting.  Production of the
	original was a Ph.D. thesis.  Excited by de Moivre's theorem?
	Take a stab at it.

	* src/roff/groff/groff.cpp (main): Comment out effect of `-J`
	option.  It remains recognized, but silently ignored (instead of
	attempting to execute a nonexistent preprocessor).
	(synopsis): Add comment reminding us to document the option if
	we ever get the support.
	(help): Comment out documentation of option.

2021-07-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	[[gnt]roff]: Make usage messages more accurate.

	* src/roff/groff/groff.cpp (synopsis): Drop `-h` and `-v` from
	normal operation synopsis.  Sort options in en_US lexicographic
	order.  Tighten presentation of alternate usage forms.
	(help): Document longer forms of `-d` and `-r`.
	* src/roff/nroff/nroff.sh: Report `$prog` instead of "nroff"
	literal.  Refer to `-P` option argument as "arg" instead of
	"opt", for consistency with groff usage message.  Use opposite
	case from option letter for option argument.
	* src/roff/troff/input.cpp (usage): Put brackets around options.
	Sort options in en_US lexicographic order.  Drop `-v` from
	normal operation synopsis.  Use opposite case from option letter
	for option argument.  Put operand name in capitals.  Set valid
	usage forms on one physical line each (see 89648fb4, 5 June).
	Add synopsis line for alternate usage forms.

2021-07-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Refactor environment initialization, switching, and
	copying.

	* src/roff/troff/env.cpp: Rename struct `env_list` to
	`env_list_node` since it describes a node of a singly-linked
	list.  Remove constant `NENVIRONMENTS` and array `env_table`.
	Add static symbol `default_environment_name` to replace string
	literal.
	(init_environments): Stop initializing `curenv` through
	`env_table`.  Use `default_environment_name` for that
	initialization and add the default environment to
	`env_dictionary`.
	(environment_switch): Simplify.  Shorten "dummy environment"
	diagnostic message.  Stop creating an integer-named
	environment inside the `env_table` array, only falling through
	to use the `env_dictionary` if the named environment is not a
	valid integer or if the array is full.  Instead use
	`env_dictionary` always.  Drop no longer needed `pop`
	quasi-Boolean integer with extra state to suppress environment
	stack underflow errors.  Instead report the error if underflow
	occurs, regardless of any other circumstance.
	(environment_copy): Simplify.  Stop searching the `env_table`
	array for an environment to copy from, only falling through to
	use the `env_dictionary` if the named environment is not a valid
	integer or if the array is full.  Instead search
	`env_dictionary` always.  Emit "no environment specified to copy
	from" diagnostic only if the `evc` request is given no argument.
	If the source environment to copy from is given but not found,
	emit a new diagnostic naming the nonexistent environment.  Fix
	bug: stop returning early if no copying could be done; instead
	fall through to the end of the function, which calls
	`skip_line()` and prevents anything on the remainder of the
	{invalid} control line from being interpreted.  Problem dates
	back to commit da3b7137, 6 March 2000 (groff 1.16).

	Fixes <https://savannah.gnu.org/bugs/?60913>.

2021-07-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/tests/evc_produces_no_output_if_invalid.sh:
	  Regression-test Savannah #60913.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

2021-07-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	Update English hyphenation patterns.

	* NEWS: Add item.
	* tmac/hyphen.en: Update file using `hyph-en-us.tex` patterns
	file from the TeX hyph-utf8 project.
	* tmac/hyphenex.en: Remove explicit hyphenations for words that
	no longer require them when using the new patterns.  Add one
	item scraped from an erratum comment in hyphen.en
	{"dem-o-crat"}.

	The new patterns likely _will_ change the automatic hyphenation
	break points of your English documents.  Here is a sample of
	affected words found within groff's own documentary corpus.

	OLD			NEW
	===			===
	ar‐range‐ment		arrange‐ment
	col‐umns		columns
	con‐struc‐ted		con‐structed
	cus‐tom‐ized		cus‐tomized
	def‐i‐ni‐tions		de‐f‐i‐n‐i‐tions
	der‐i‐va‐tions		de‐riva‐tions
	hy‐phen‐a‐tion		hy‐phen‐ation
	ma‐te‐rial		ma‐te‐r‐ial
	Mi‐cro‐soft		Mi‐crosoft
	pipe‐lines		pipelines
	post‐pro‐ces‐sors	post‐proces‐sors
	pro‐cessed		processed
	pro‐cesses		processes
	spa‐ces			spaces
	Wer‐ner			Werner

2021-07-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Add unit test for `substring` request.

	* src/roff/groff/tests/substring_works.sh: Do it.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

2021-07-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	[tmac]: Remove compatibility mode save registers after using
	them.

	* tmac/an.tmac:
	* tmac/andoc.tmac:
	* tmac/cp1047.tmac:
	* tmac/devtag.tmac:
	* tmac/ec.tmac:
	* tmac/fallbacks.tmac:
	* tmac/latin1.tmac:
	* tmac/latin2.tmac:
	* tmac/latin5.tmac:
	* tmac/latin9.tmac:
	* tmac/papersize.tmac:
	* tmac/pdfpic.tmac:
	* tmac/psold.tmac:
	* tmac/pspic.tmac:
	* tmac/trace.tmac:
	* tmac/unicode.tmac: Do it.

	* doc/groff.texi (Implementation Differences):
	* man/groff_diff.7.man (Implementation Differences): Illustrate
	doing so in relevant examples.

2021-07-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/troffrc: Load the English localization file `en.tmac` by
	default, instead of dealing with hyphenation language and
	patterns directly in this file.  (This also replaces a
	localization-determination mechanism that was never part of a
	groff release.)

	* src/roff/groff/tests/initialization_is_quiet.sh:
	* src/roff/groff/tests/localization_works.sh: Rewrite to use
	explicit macro file loading localization mechanism.

2021-07-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	Move core man(7) implementation to an.tmac.

	* tmac/an.tmac: Delete andoc-sourcing wrapper.

	* tmac/an-old.tmac: Rename...
	* tmac/an.tmac: ...to this.

	* tmac/tests/an-old_AT-and-UC-footer-saved-and-restored.sh:
	* tmac/tests/an-old_CS-register-off.sh:
	* tmac/tests/an-old_CS-register-on.sh:
	* tmac/tests/an-old_CS-register-unspecified.sh:
	* tmac/tests/an-old_CT-register-off.sh:
	* tmac/tests/an-old_CT-register-on.sh:
	* tmac/tests/an-old_CT-register-unspecified.sh:
	* tmac/tests/an-old_FT-bad-value-should-not-trash-titles.sh:
	* tmac/tests/an-old_LL-init-sanely.sh:
	* tmac/tests/an-old_TH-repairs-ad-damage.sh:
	* tmac/tests/an-old_TH-repairs-hy-damage.sh:
	* tmac/tests/an-old_TS-do-not-keep-tables-when-cR-set.sh:
	* tmac/tests/an-old_X-register-works.sh:
	* tmac/tests/an-old_avoid-two-font-denial-of-service.sh:
	* tmac/tests/an-old_no-break-after-short-paragraph-tags.sh:
	* tmac/tests/an-old_page-footers-present.sh:
	* tmac/tests/an-old_page-header-has-current-data.sh:
	* tmac/tests/an-old_title-abbreviation-works.sh: Rename...

	* tmac/tests/an_AT-and-UC-footer-saved-and-restored.sh:
	* tmac/tests/an_CS-register-off.sh:
	* tmac/tests/an_CS-register-on.sh:
	* tmac/tests/an_CS-register-unspecified.sh:
	* tmac/tests/an_CT-register-off.sh:
	* tmac/tests/an_CT-register-on.sh:
	* tmac/tests/an_CT-register-unspecified.sh:
	* tmac/tests/an_FT-bad-value-should-not-trash-titles.sh:
	* tmac/tests/an_LL-init-sanely.sh:
	* tmac/tests/an_TH-repairs-ad-damage.sh:
	* tmac/tests/an_TH-repairs-hy-damage.sh:
	* tmac/tests/an_TS-do-not-keep-tables-when-cR-set.sh:
	* tmac/tests/an_X-register-works.sh:
	* tmac/tests/an_avoid-two-font-denial-of-service.sh:
	* tmac/tests/an_no-break-after-short-paragraph-tags.sh:
	* tmac/tests/an_page-footers-present.sh:
	* tmac/tests/an_page-header-has-current-data.sh:
	* tmac/tests/an_title-abbreviation-works.sh: ...to these.

	* tmac/tmac.am (TMACNORMALFILES): Remove an-old.tmac.
	(tmac_TESTS): Reflect renames of test files above.
	(tmac/stamp-wrap): Macro-source `an.tmac` in the man wrapper.

	* tmac/andoc.tmac: Refer to (and macro-source) an.tmac.
	* tmac/man.tmac: Macro-source `an.tmac`, not `andoc.tmac`.
	* tmac/tests/andoc_flush-between-packages.sh: Call groff with
	`-mandoc` option instead of `-man`.  Henceforth, if you want the
	andoc wrapper, you have to ask for it.

	* tmac/an.tmac: Drop "-old" nomenclature.

	* PROBLEMS:
	* doc/groff.texi (man):
	* tmac/man.local: Update documentation appropriately.

	* NEWS: Add item.

	* man/groff_tmac.5.man (Macro packages/man pages):
	* src/roff/groff/groff.1.man (Usage/Macro packages): Update
	discussion of an, doc, and andoc.
	* tmac/groff_man.7.man.in (Files): Update descriptions of
	an.tmac, andoc.tmac, man.tmac, and mandoc.tmac.

	Fixes <https://savannah.gnu.org/bugs/?60789>.

2021-07-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Slightly refactor.  Make the source code slightly more
	accessible by using a `bool` type for a function's default
	parameter used as a Boolean, and comment it at call sites where
	the default is overridden.

	* src/roff/troff/token.h (get_name, get_long_name): Change type
	of parameter from `int` to `bool` and default from 0 to `false`.
	* src/roff/troff/input.cpp (do_get_long_name,
	empty_name_warning, get_name, get_long_name): Change type of
	parameter from `int` to `bool` in declarations and definitions.
	* src/roff/troff/column.cpp (column_justify):
	* src/roff/troff/div.cpp (diversion_trap):
	* src/roff/troff/env.cpp (environment_switch, environment_copy,
	do_input_trap, set_hyphenation_language,
	do_hyphenation_patterns_file):
	* src/roff/troff/input.cpp (define_color, composite_request,
	do_define_string, do_define_macro, rename_macro, alias_macro,
	chop_macro, do_string_case_transform, substring_request,
	length_request, asciify_macro, unformat_macro, do_register,
	device_macro_request, do_if_request, do_source, ps_bbox_request,
	do_open, close_request, do_write_request, write_macro_request,
	define_class, copy_file, vjustify, transparent_file,
	do_macro_source):
	* src/roff/troff/node.cpp (font_translate, font_position, style,
	get_fontno):
	* src/roff/troff/reg.cpp (define_number_reg, inline_define_reg,
	alter_format, alias_reg, rename_reg): Update call sites.  Add
	comment indicating meaning of parameter.

	Do similarly for token::delimiter() member function.

	* src/roff/troff/token.h (token::delimiter): Change type of
	parameter from `int` to `bool`, name from `warn` to `err` (to
	match definition and behavior in input.cpp) and default from 0
	to `false`.
	* src/roff/troff/input.cpp (token::delimiter): Change type of
	parameter from `int` to `bool` in definition.
	* src/roff/troff/input.cpp (do_expr_test, get_delim_number,
	get_line_arg, read_size, do_register, read_draw_node):
	* src/roff/troff/reg.cpp (inline_define_reg): Update call sites.
	Add comment indicating meaning of parameter.

2021-07-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-footer): Remove string
	`an-outer-footer-text` after we're done with it.

2021-07-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	Remove the stripper script and its vestiges.

	* contrib/hdtbl/hdmisc.tmac:
	* contrib/hdtbl/hdtbl.tmac:
	* contrib/mom/om.tmac:
	* tmac/doc.tmac:
	* tmac/e.tmac:
	* tmac/mdoc/doc-common:
	* tmac/mdoc/doc-ditroff:
	* tmac/mdoc/doc-nroff:
	* tmac/mdoc/doc-syms: Remove `%beginstrip` comment.

	* tmac/strip.sed: Delete.

	* tmac/tmac.am (EXTRA_DIST): Stop shipping `strip.sed`.

	Fixes <https://savannah.gnu.org/bugs/?55091>.

2021-07-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	Skip the stripper, part 3/3 (mdoc).

	* tmac/doc-old.tmac-u:
	* tmac/doc.tmac-u:
	* tmac/mdoc/doc-common-u:
	* tmac/mdoc/doc-ditroff-u:
	* tmac/mdoc/doc-nroff-u:
	* tmac/mdoc/doc-syms-u: Rename these...

	* tmac/doc-old.tmac:
	* tmac/doc.tmac:
	* tmac/mdoc/doc-common:
	* tmac/mdoc/doc-ditroff:
	* tmac/mdoc/doc-nroff:
	* tmac/mdoc/doc-syms: ...to these.

	* tmac/tmac.am (TMACNORMALFILES): Add `doc-old.tmac` and
	`doc.tmac`.
	(TMACUNSTRIPFILES, TMACSTRIPFILES): Delete variables.
	(dist_tmac_DATA): Drop `$(TMACUNSTRIPFILES)`.
	(nodist_tmac_DATA): Drop `$(TMACSTRIPFILES)`.
	(TMACMDOCSTRIPFILES): Rename...
	(TMACMDOCFILES): ...to this.
	(nodist_mdoc_DATA, MOSTLYCLEANFILES): Reflect above rename.
	(nodist_mdoc_DATA): Rename...
	(dist_mdoc_DATA): ...to this.
	(TMACMDOCUNSTRIPFILES): Delete variable.
	(EXTRA_DIST): Drop `$(TMACMDOCUNSTRIPFILES)`.
	($(TMACSTRIPFILES), $(TMACMDOCSTRIPFILES)): Drop targets.

	* tmac/doc.tmac:
	* tmac/mdoc/doc-common:
	* tmac/mdoc/doc-ditroff:
	* tmac/mdoc/doc-nroff:
	* tmac/mdoc/doc-syms: Bracket macro definitions tightly with
	`eo` and `ec` requests since they were written under this
	assumption.  The files in general were also drafted under the
	assumption that they'd be stripped, so `eo` was in effect for
	broad swaths of the file including portions outside of any macro
	definition where comments were used.  But comments are expressed
	using the escape character!  So we got warnings about an
	undefined '\"' macro, and that was just the beginning of a
	wrecking ball of trouble that smashed through the
	implementation.  Some--but not all--string definitions (and
	appendments) needed to be bracketed with `eo` and `ec` as well,
	to delay interpolation of embedded string names that are not
	defined until macros interpolate them later.  I did what
	appeared necessary to pass our tests and keep -Tutf8 and -Tps
	renderings of tmac/groff_mdoc.7 from changing from their
	strip-influenced output.  Subtle bugs might linger, but some
	won't be the fault of the unstripping process.  (For instance,
	try provoking a usage message on the `Hf` macro.)

2021-07-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man pages]: Remove compatibility save register.

	* **/*.man*: Remove page-local compatibility save register after
	using it.

2021-07-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	[tmac]: Simplify `mso` requests.

	* tmac/X.tmac:
	* tmac/Xps.tmac:
	* tmac/dvi.tmac:
	* tmac/html.tmac:
	* tmac/lbp.tmac:
	* tmac/lj4.tmac:
	* tmac/ps.tmac:
	* tmac/tty-char.tmac:
	* tmac/tty.tmac:
	* tmac/www.tmac.in: Move `mso` requests to be inside regions
	where compatibility mode is turned off.  I believe they were
	placed where they were because things didn't work when the
	compatibility mode register being saved was .C, rather than the
	.cp register introduced in commit 6a37bb5f, 17 April 2020.  Now
	the logic is less mysterious.  Also, remove the register used to
	save compatibility mode after we're done with it, to reduce name
	space clutter.

2021-07-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/it.tmac: Drop sourcing of `latin1.tmac` and set up of
	hyphenation codes; since the `hyphen.it` file is pure ASCII, we
	don't need to expect Latin-1-encoded code points in it.

2021-07-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	[tmac]: Make localization files work in compatibility mode.

	* tmac/cs.tmac:
	* tmac/de.tmac:
	* tmac/en.tmac:
	* tmac/fr.tmac:
	* tmac/it.tmac:
	* tmac/ja.tmac:
	* tmac/sv.tmac:
	* tmac/zh.tmac: Save and restore compatibility mode.

	* tmac/den.tmac: Invoke requests with `do`.

	Fixes <https://savannah.gnu.org/bugs/?60874>.  Thanks to Bjarni
	Ingi Gislason for the report.

2021-07-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #60874.

	* src/roff/groff/tests/initialization_is_quiet.sh: Test it.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

2021-07-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	Integrate Italian localization.

	* NEWS: Add item.  Update an existing list of localization macro
	files.
	* doc/groff.texi (Manipulating Hyphenation):  Update table of
	hyphenation pattern left and right minimums and list of
	available localization files to include Italian.
	* man/groff_tmac.5.man: Update list of available localization
	files to include Italian.
	* src/roff/groff/tests/localization_works.sh: Test it.
	* tmac/tmac.am (TMACNORMALFILES): Add `it.tmac` and `hyphen.it`.

	Fixes <https://savannah.gnu.org/bugs/?60828>.

2021-07-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	[tmac]: Flesh out Italian localization.

	* tmac/hyphen.it: Add hyphenation patterns from TeX hyph-utf8
	project.
	* tmac/it.tmac:
	  - Revise copyright date to just this year (it had been copied
	    from fr.tmac).
	  - Use grave accents instead of acute ones to indicate stress
	    on word-final vowels.  Per the University of Wikipedia and a
	    sampling of apparently well-typeset Italian language
	    documents online, this appears to be correct.  Yell at me if
	    I'm wrong.
	  - Source latin1.tmac instead of latin9.tmac; Italian doesn't
	    require any letter code points from Latin-9, unlike the
	    French file upon which this one was based, and Latin-1 is
	    less of a pain to work with.
	  - Set the hyphenation mode to 1 instead of 4 for congruence
	    with the requirements of the hyphenation patterns.
	  - Set up hyphenation codes.  This (and sourcing latin1.tmac)
	    might not be necessary as the hyphenation pattern file, of a
	    more recent vintage than the others we're using, is pure
	    ASCII.  However I don't understand these issues perfectly so
	    I am making the conservative choice.  Again per the
	    University of Wikipedia, Italian requires only acute and
	    grave accents, in contrast to the rich French diacritical
	    bouillabaisse.
	  - Set the hyphenation language and load the pattern file.
	  - Add comments in parallel with other localization files and
	    to supply text editors with hints.

2021-07-02  Edmond Orignac <edmond.orignac@wanadoo.fr>

	* tmac/it.tmac: Add Italian localization.

2021-07-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	Skip the stripper, part 1/3 ("me").

	* tmac/e.tmac-u: Rename to...
	* tmac/e.tmac: ...this.

	* tmac/tmac.am (TMACNORMALFILES): Add tmac/e.tmac.
	(TMACUNSTRIPFILES): Remove tmac/e.tmac-u.
	(TMACSTRIPFILES): Remove tmac/e.tmac.

	Fixes one third of <https://savannah.gnu.org/bugs/?55091>.

2021-07-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.pl: Simplify parsing.  Dave Kemper pointed
	out that preprocessors like pic(1) use pretty unsophisticated
	*roff parsing to determine where to perform their textual
	replacements.  My enhancements to support input line
	continuation and cope with brace escapes were thus
	overengineered.  Remove them.
	  - Drop scalars `is_continued_line` and `logical_line`.
	  (do_line): Stop performing logical line concatenation and
	  detecting input line continuation.  Perform operations on
	  `line` instead of `logical_line`.  Stop removing brace
	  escapes.
	* src/utils/grog/grog.1.man (Limitations): Update discussion.

	Fixes <https://savannah.gnu.org/bugs/?60862>.  Thanks, Dave!

2021-06-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/reg.cpp (lookup_number_reg, alias_reg): In
	diagnostic messages, say simply "register" instead of "number
	register" (there is no other kind).

2021-06-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grog]: Refactor input parsing.

	* src/utils/grog/grog.pl:
	  - Add scalar `use_compatibility_mode` (see below).
	  - Add list `request` to store the names of all requests
	    recognized by groff so that they aren't confused with macro
	    names.
	  - Add scalars `have_seen_first_macro_call` (replaces
	    `before_first_command`, but at global scope),
	    `is_continued_line` and `logical_line`.  The latter two
	    enable us to handle *roff input line continuation correctly.
	  (process_arguments): Set `use_compatibility_mode` if `-C`
	  option specified.
	  (process_input): Refactor to greatly simplify, to not attempt
	  to read the first line of an input file as a special case, and
	  to avoid sending `do_line` an undefined argument (when EOF is
	  reached).
	  (do_first_line): Delete.
	  (do_line): Rewrite the early stages of input parsing.
	  - Concatenate continued input lines, setting
	    `is_continued_line` and returning early as each one is seen,
	    storing the accumulating input in `logical_line`.
	  - Check the input line for the form of comment deposited by
	    Perl's Pod::Man, which uses a highly accented dialect of
	    man(7); if it's present, inflate `man_score` to compensate
	    for the page-private `IX` macro it defines but which
	    duplicates the name and function of a 4.2BSD-era ms(7)
	    extension that would otherwise deceive our scoring
	    mechanism, because Pod::Man produces `IX` calls to
	    metastatic excess.  (An alternative to this kludge is
	    documented in comments: if a "standard" macro is redefined,
	    we could delete it from the relevant lists and hashes.)
	  - Strip *roff comments from input.
	  - Normalize control lines; convert the no-break control
	    character to the regular one and remove unnecessary
	    white space.
	  - Remove brace escapes.
	  - Recognize two-character macro calls when not followed by
	    white space in compatibility mode.
	  - Drop logic that erroneously attempted to infer soelim(1) use
	    from macro calls and request invocations.  The grog(1) and
	    soelim(1) man pages now both explain why such an effort was
	    misguided.
	  - Recognize macro definitions created by .am and .am1 requests
	    {not just .de and .de1}.
	  - Ignore all other *roff requests.
	  - What remains must be a ("standard") macro call, so set
	    `have_seen_first_macro_call`.

	* src/utils/grog/grog.1.man (Limitations): Document a further
	restriction: don't change the escape character, either.

	* src/utils/grog/tests/smoke-test.sh: Comment out pic-detection
	test on soelim(1).  The pic macro calls are guarded by roff
	control structures and only worked previously by accident
	because grog did not recognize *roff input line continuation,
	now it does and the illusion is dispelled.  (A reliable way to
	fool grog before and after my refactoring is now documented in
	its man page.)

	Fixes <https://savannah.gnu.org/bugs/?59622>.

2021-06-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #59622.

	* src/utils/grog/tests/recognize-perl-pod.sh: Test it.
	* src/utils/grog/grog.am (grog_TESTS): Run test.

2021-06-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.1.man (Limitations): Document further
	restrictions (no changing of control characters, control
	structures not interpreted).

2021-06-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grog]: Refactor command-line argument handling.

	* src/utils/grog/grog.pl:
	  - Drop scalars `groff_opts`, `device`, and `with_warnings`.
	  - Move scalar `pdf_with_ligatures`...
	  (process_arguments): ...here.  Recognize all groff options
	  that take an optionally whitespace-separated option argument
	  and apply the existing "delayed option" handling to them.
	  Push any groff option except `-m` as-is onto the constructed
	  groff command's argument list (`-m` handling is unchanged).
	  Match long option names exactly, not sloppily.  Stop
	  recognizing `--warnings` option.  Emit error diagnostic if
	  unrecognized long option encountered.
	  (infer_device): Delete subroutine and its top-level call site.
	  (help): Undocument `--warnings`.

	* src/utils/grog/grog.1.man:
	  (Synopsis): Use font style macros instead of .OP.  Undocument
	  `--warnings`.
	  (Options): Note that the groff arguments produced by the grog
	  `--ligatures` option are supported only by the `pdf` device.
	  Undocument `--warnings`.  Undocument groff-incompatible
	  restriction on whitespace before option arguments; grog is
	  compatible now.  Simplify discussion in light of simplified
	  grog logic.
	  (Details): Undocument scenario where grog infers multiple
	  main/major/full-service macro packages; this no longer
	  happens.
	  (Examples): Update to no longer illustrate inclusion of `-T
	  ps` option; this no longer happens by default.

	* NEWS: Document removal of grog's `--warnings` option.

	Fixes <https://savannah.gnu.org/bugs/?57873>; groff options are
	no longer "mangled".

2021-06-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #57873.

	* src/utils/grog/tests/preserve-groff-options.sh: Test it.
	* src/utils/grog/grog.am (grog_TESTS): Run test.

2021-06-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grog]: Refactor device (`-T` option) handling.

	* src/utils/grog/grog.pl: Redeclare `device` as a scalar instead
	of a list.
	(process_arguments): Generalize handling of `-T` and `-m`
	options to permit optional whitespace.  Rename scalar `was_T` to
	`delayed_option`.
	(infer_device): Stop unconditionally adding a `-T ps` argument
	to the generated groff command.  Remove logic that validates any
	given `-T` option argument; instead, let groff fail if a bad one
	is supplied (its own diagnostics in this scenario were improved
	in commit 5a721a30, 27 May).  This makes grog agnostic about any
	differently configured default device in groff itself, and about
	the GROFF_TYPESETTER environment variable.

	Fixes <https://savannah.gnu.org/bugs/?55301> by getting out of
	the way.

	* src/utils/grog/tests/PF-does-not-start-pic-region.sh:
	* src/utils/grog/tests/smoke-test.sh: Update expected output.

2021-06-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.am (grog): Drop sed replacement of unused
	configuration variables @g@, @BINDIR@, @libdir@, and @EGREP@.
	* src/utils/grog/grog.pl (process_arguments, construct_command):
	Tweak warning diagnostics to use a style where an empty argument
	will be easy to spot.
	(do_first_line): Stop emitting pseudo-diagnostic (to standard
	output!) if we see a "groff options line" [it's not] of
	incorrect form.  The preprocessor encoding hint convention of
	concern here is particular to man pages, disregarded by at least
	one major implementation thereof (man-db man), and not used by
	other types of roff documents; at this point, we don't yet know
	if the document we're processing is a man page.
	(construct_command): Correct comment and related logic error.
	Clear the inferred main macro package if _any_ `-m` argument
	matching any main macro package is given.  Warn only if what is
	specified doesn't match the inference.

2021-06-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grog]: Heavily refactor.

	* src/utils/grog/grog.pl:
	  - Drop import of unused module `Data::Dumper`.
	  - Drop unused scalars `Sp` and `correct_tmac`.
	  - Simplify determination of version number.  Drop hash `at_at`
	    which only stored one key, `GROFF_VERSION`.  Initialize
	    scalar `groff_version` to "DEVELOPMENT".  Rename scalar
	    `before_make` to `in_source_tree` and initialize to zero.
	    Update `groff_version` with Automake-determined version
	    variable if it is defined (i.e., grog is not running outside
	    of, or in an unbuilt, groff source or binary distribution).
	  - Drop unused `Mparams` list.  Replace it with new list
	    `requested_package`, which stores the arguments to any grog
	    `-m` options specified by the user.
	  - Rename many objects so that I, and others, can better
	    comprehend their purpose, and for consistent letter casing.
	    . @Command -> @command
	    . @devices -> @device
	    . $Prog -> $program_name
	    . %macros -> %user_macro
	    . $have_any_valid_args -> $have_any_valid_arguments
	    . &handle_args -> &process_arguments
	    . &handle_whole_files -> &process_input
	    . @preprograms -> @preprocessor
	    . &make_groff_device -> &infer_device
	    . &make_groff_preproc -> &infer_preprocessors
	    . &make_groff_tmac_others -> &infer_macro_packages
	    . &make_groff_tmac_man_ms -> &infer_man_or_ms_package
	    . &make_groff_line_rest -> &construct_command
	  - Drop many unused keys in `Groff` hash.
	  - Add new lists, `macro_ms`, `macro_man`, and
	    `macro_man_or_ms` to support new scoring technique to
	    disambiguate input documents between these two packages.
	  - Append the foregoing 3 lists to new list `standard_macro`,
	    and add these as keys to the `Groff` hash.
	  - Add new list `main_package` to keep track of full-service
	    package names.
	  - Add new scalars `man_score`, `ms_score`, and
	    `inside_tbl_table` to aid disambiguation of .TH macro calls
	    and the many macro names shared between man(7) and ms(7).
	  (process_arguments): Strip '-m' off of argument before storing
	  the remainder in `@requested_package`.
	  (do_line): Detect .TH macro call even if white space occurs
	  between the control character and the macro name.
	  (do_line): Inflate `$man_score` by 100 if .TH is the first
	  macro call seen in a document.
	  (do_line): Fix bug; clear `$before_first_command` in correct
	  scope--after any macro call, not just if we saw a .TH as the
	  first macro call.
	  (do_line): Set `$inside_tbl_table` when we see a .TS call.
	  (do_line): Clear `$inside_tbl_table` when we see a .TE call.
	  Also increment `$Groff{'tbl')' again, increasing the "score"
	  of tbl(1) usage evidence.
	  (do_line): Drop a lot of code that manually increments %Groff
	  keys corresponding to man and ms macros.  This is now done
	  differently and elsewhere.
	  (do_line): Drop "P" from list of characteristic mm(7) macros.
	  (do_line): Simplify matching of mom(7) macros (match $command,
	  not $line).  Extend list of characteristic mom(7) macros.
	  (do_line): Increment $Groff{$key} if $key is in
	  @standard_macro.
	  (infer_man_or_ms_package): Rewrite.  Compute a score for each
	  package by counting occurrences of their characteristic
	  macros.  If both have a score of zero, assume that the input
	  is a raw roff document.  If the scores are equal
	  {doc/webpage.ms, startlingly, comes within 1 point of a tied
	  score}, infer ms(7) if 'TH' was never called, and if it was,
	  issue a diagnostic advising user to supply a disambiguating
	  `-m` option.  Otherwise, the scores are unequal: infer the
	  package of the winner.  Set scalar `inferred_main_package`
	  instead of pushing `-m` options onto `@m`.
	  (infer_macro_packages): Set scalar `inferred_main_package`
	  instead of pushing `-m` options onto `@m`.  Explicitly return
	  0 if we fall off the end of the function.
	  (construct_command): Rewrite handling of -m options.  Add new
	  list `msupp` to store supplementary (non-main) macro package
	  arguments.  If a full-service package was explicitly
	  requested, it had better not clash with what we inferred.  If
	  it does, explicitly unset `inferred_main_package` so that the
	  -m arguments are placed in the same order that the user gave
	  them; caveat dictator.  If `--run` option was given, just
	  print the command; don't preface it with __FILE__ and __LINE__
	  noise.
	  - Remove comments documenting shared variables used by
	    subroutines.  These are far from useless but too tedious to
	    keep up to date while the code is in flux.
	  - Note several places for further code review or refactoring
	    with "XXX" comments.
	  - Add Vim modeline.

	grog now passes all its tests and correctly infers arguments for
	all in-tree groff documents (except for a known, and already
	documented in grog(1), false positive detection of soelim in
	soelim(1)).  This refactor also obviates or resolves several
	outstanding Savannah tickets.

	Fixes <https://savannah.gnu.org/bugs/?44707> by obviating it;
	grog no longer cares about file name extensions on man pages (or
	any other input).

	Fixes <https://savannah.gnu.org/bugs/?55302>; same.  The quality
	of diagnostic messages has been improved as well.

	Fixes <https://savannah.gnu.org/bugs/?59753>; same.

	Fixes <https://savannah.gnu.org/bugs/?59664>.  The attached
	patch was a less aggressive refactor of &do_line and %Groff.
	Its author made the following claim for it: "With this patch,
	all 'man', 'me', 'mom, and 'ms' files in the repository are
	correctly identified.  The only example of a 'mm'-file is
	"letter.mm", which is not recognized correctly."  As noted
	above, the present refactor achieves correct recognition of all
	of the files, including letter.mm.

2021-06-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/tests/smoke-test.sh: Perform whole-line
	matches.  Apply DRY principle to expected output.  In
	anticipation of pending changes to grog.pl, uncomment and add
	tests for several in-tree documents.
		doc/meref.me
		contrib/mom/examples/copyright-chapter.mom
		contrib/mom/examples/copyright-default.mom
		contrib/mom/examples/letter.mom
		contrib/mom/examples/mom-pdf.mom
		contrib/mom/examples/mon_premier_doc.mom
		contrib/mom/examples/sample_docs.mom
		contrib/mom/examples/slide-demo.mom
		contrib/mom/examples/typesetting.mom
		doc/webpage.ms

2021-06-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grog]: Refactor.  Discard filename extension-based inference.

	* src/utils/grog/grog.pl (do_line): Stop saving the control
	character in the matched scalar `command` (the request or macro
	name).  We don't need it, and it simplifies much later matching.
	(handle_file_ext): Delete.
	(top level): Drop call site.

	* src/utils/grog/grog.pl: Rename `tmac_ext` scalar to
	`inferred_main_package`.

2021-06-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.pl (handle_args): Complain less noisily
	and more comprehensibly when given unrecognized groff options.

2021-06-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grog]: Handle "--" argument as documented.

	* src/utils/grog/grog.pl (handle_args): Rename scalar
	`double_minus` to `no_more_options` so it actually communicates
	something.  Drop openability check of operand encountered after
	"--"; since it's on the operand list, it will be checked later
	{in `handle_whole_files`}.  Stop pushing "--" itself onto the
	`filespec` list.

	Fixes <https://savannah.gnu.org/bugs/index.php?60834>.

2021-06-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grog]: Revise operand handling, diagnostics, and exit status.

	* src/utils/grog/grog.pl: Track more state so that we can
	process the argument list more intelligently and exit with a
	meaningful status.  Add `had_inference_problem`,
	`had_processing_problem`, and `have_any_valid_args` Boolean
	scalars.
	(fail): Add diagnostic subroutine for serious problems; sets
	`had_processing_problem`.
	(handle_args, handle_file_ext): Stop complaining here about
	unopenable file operands.
	(handle_whole_files): Complain only here, with `fail` if a file
	operand cannot be opened.
	(make_groff_line_rest): Don't exit immediately if there is a
	macro package inference clash, because there might be more
	operands to process.  Instead, set `had_inference_problem`.
	Stop exiting from this subroutine.
	(top level): Only call the inference subroutines if we had a
	valid operand to work with (including an implicit read from
	stdin).  Exit with a status corresponding to the Booleans
	declared above.

	* src/utils/grog/grog.1.man (Exit status): Add section; describe
	semantics of exit status values.

	Fixes <https://savannah.gnu.org/bugs/index.php?60833>.

2021-06-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grog]: Revise diagnostic messages.

	* src/utils/grog/grog.1.man (err): Rename from this...
	(warn): ...to this.  Prefix diagnostic with command name and
	diagnostic severity level.  The subroutine didn't change the
	exit status and was not used to report serious trouble, so
	"warning" seems appropriate.
	(make_groff_tmac_man_ms): Update call sites.  Modify diagnostic
	wording to stop calling macros "requests".

2021-06-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/grog.1.man (Limitations): Add subsection
	covering the problem grog has with soelim inference.  Undocument
	groff's -s option as one that can be inferred, because at
	present it cannot be.

	Prompted by Savannah #60421.

2021-06-26  Dave Kemper <saint.snit@gmail.com>

	[grog]: Fix erroneous detection of pic(1) usage.

	Fixes <https://savannah.gnu.org/bugs/?60772>.

2021-06-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/grog/tests/PF-does-not-start-pic-region.sh:
	  Regression-test Savannah #60772.
	* src/utils/grog/grog.am (grog_TESTS): Run test.

2021-06-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grog]: Add smoke test.

	* src/utils/grog/tests/smoke-test.sh: Add test.
	* src/utils/grog/grog.am (grog_TESTS): Run it.
	(TESTS): Add `grog_TESTS`.
	(EXTRA_DIST): Ship test.

2021-06-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grog]: Relocate in source tree.

	* src/roff/grog: Move from here...
	* src/utils/grog: ...to here.

	* MANIFEST:
	* Makefile.am:
	* src/utils/grog/grog.am: Reflect move.

	Fixes <https://savannah.gnu.org/bugs/?60788>.

2021-06-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Assigning Formats): Revise discussion.
	Clarify that a format never causes truncation of an interpolated
	magnitude.  Note that non-Arabic number formats cannot be used
	as operands to arithmetic expressions.  Add example of how to
	work around this.

2021-06-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Clarify diagnostics.

	* src/roff/troff/reg.cpp (reg::set_increment): Update diagnostic
	to unabbreviate "auto".
	(reg::alter_format): Update diagnostic to say "assign" instead
	of "alter", since the documentation consistently uses the former
	term.

2021-06-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	[troff]: Slightly refactor.  Make the source code slightly more
	accessible by using a `bool` type for a member function's
	default parameter used as a Boolean, and comment it at call
	sites where the default is overridden.

	* src/roff/troff/token.h (token::get_char): Change type of
	parameter from `int` to `bool` and default from 0 to `false`.
	* src/roff/troff/input.cpp (token::get_char): Update definition.
	* src/roff/troff/env.cpp (hyphen_word):
	* src/roff/troff/node.cpp (remove_font_special_character):
	* src/roff/troff/input.cpp (do_overstrike, do_bracket, next,
	do_define_character, remove_character, get_line_arg,
	encode_char, do_if_request, do_translate, char_flags,
	hyphenation_code, define_class): Update call sites.  Add comment
	indicating meaning of parameter.

2021-06-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (TH): When outputting HTML, write the
	possibly case-transformed (but not yet abbreviated) page title
	as the `title` element instead of the unaltered first argument
	to `TH`.

2021-06-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-abbreviate-title): Define with `de`, not
	`de1`, since it is not intended for call by traps or users.

2021-06-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Tab Stops): Fix erroneous claim; tab stops are
	not set "every half inch across the page".  This is only true in
	"line tabs mode", which is not enabled by default.  The default
	behavior, to keep compatibility with AT&T troff, measures tab
	stops relative to the current position on the input line.

2021-06-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/mdoc/doc-nroff-u: Set the usekeeps register used by tbl
	to the logical complement of the cR (continuous rendering)
	register.  This prevents blank lines from creeping into tables
	in that mode.

2021-06-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-blank-line-trap, an-leading-space-trap):
	We say in a comment, "Macros ... that are called by traps of any
	kind must be defined with `de1` because they might be called
	from a context where compatibility mode is enabled."  So do it.

2021-06-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-header): Lightly refactor; tighten.

2021-06-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-write-paragraph-tag): Lightly refactor
	for clarity and to reduce code duplicated in both branches of an
	`ie`/`el` structure.

2021-06-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Rename `an-HF-remap-I-style` register to
	`an-remap-I-style-in-headings` for greater clarity.

2021-06-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/mdoc/doc-common-u (doc-header, doc-end-macro): Partially
	revert bf4b3dde (from 20 May); per Ingo Schwarze, continuous
	rendering in mdoc(7) documents has always used only one vee
	between headers, footers, and man page body text.

2021-06-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/en.tmac: Stop using `do` request; it's unnecessary since
	the file is already loaded with compatibility mode off.

2021-06-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	[ms]: Fix doc omission; .ID can take an argument.

	* doc/groff.texi (ms Document Control Settings) <DI>:
	Characterize register setting as a _default_.
	(ms Displays and Keeps): Add optional variable "indent" argument
	to syntax summary of ".DS I" and ".ID".  Note that the given
	indentation is used if present, and \n[DI] otherwise.
	* doc/ms.ms (Displays and keeps): Add optional variable "indent"
	argument to syntax summary of ".ID".  Switch fonts more
	carefully and apply italic corrections.
	* tmac/groff_ms.7.man (Displays and keeps): Sync with doc/ms.ms.

2021-06-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac: Properly situate devtag flag-related register
	names in the package name space.  Rename `need_eo_h` to
	`s@devtag-needs-end-of-heading'.  Rename `need_eo_tl` to
	`s@devtag-needs-end-of-title`.
	((initialization), par@reset, @SH, par@finish, @NH): Do it.
	(par@reset): Test their values more idiomatically.

	Fixes last third of <https://savannah.gnu.org/bugs/?42515>.

2021-06-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/e.tmac-u (@h): Migrate SCCS revision 2.37 to be idiomatic
	for groff (and not throw warnings of type 'mac').

2021-06-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/doc.am: Improve diagnostics during documentation
	generation; none should occur normally, but when they do, seeing
	troff complain about "<standard input>" is disheartening.
	(DOC_GROFF_ONLY): Add -b option so we get backtraces (also kill
	off trailing whitespace).
	(DOC_GROFF):
	(doc/pic.html):
	(doc/webpage.html): Add sed expression to inject `lf` request
	into the stream to identify the file being processed.

2021-06-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/me-revisions: Supplement history with subsequent BSD work
	from versions 2.29 (1988-04-22) to 8.1 (1993-06-05).

2021-06-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/me-revisions: Add 'me' macro package revision log, scraped
	from <https://svnweb.freebsd.org/csrg/share/me/revisions>.
	* doc/doc.am (DOCFILES): Ship it.

2021-06-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/e.tmac-u: Rename registers for clarity.  Unlike other
	macro packages supplied by groff, "me" mostly remains within the
	AT&T 2-character name space, accessing extended names
	selectively through `do` and `de1` requests.  The package
	therefore has no characteristic name prefix.
	  need_eo_h  -> devtag-needs-end-of-heading
	  need_eo_tl -> devtag-needs-end-of-title
	  need_tl    -> devtag-needs-title
	(@html_check_need_title, sh, EQ): The foregoing registers are
	effectively Booleans, so perform Boolean-style tests on their
	interpolations.

	Fixes a second third of <https://savannah.gnu.org/bugs/?42515>.

2021-06-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/e.tmac-u: Work in compatibility mode.  It seems this was
	the intention (reasonable, since "me" originates in pre-groff
	BSD Unix), but there were several problems.
	(@R): Prefix register existence test with `do`.
	(@html_check_need_title): Define with `do` since the macro name
	is long, and also with `de1` so that we can freely use groff
	extensions within.  Drop many `do` requests from macro
	definition.
	(@check_need_title): Define with `do`, as above.  Add comment
	explaining why `de1` is not needed for this macro definition.
	(@S): Prefix macro/string existence test with `do`.  Remove
	spurious space so that the argument string name is truly defined
	as empty.
	(nm, n1, n2): Prefix handling of long register names `_#p` and
	`_#f` with `do` request.
	(top level): Prefix `mso` request with `do`.
	((x-html): Define with `do` and `de1` instead of `de`.  Drop
	now-unnecessary `do` requests from macro body.
	()x-html): Define with `do` and `de1` instead of `de` for
	symmetry.
	(xp-html): Define with `do` and `de1` instead of `de`.  Drop
	now-unnecessary `do` requests from macro body.

	Fixes <https://savannah.gnu.org/bugs/?60785>.

2021-06-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/devtag.tmac: Work in compatibility mode; define macros
	with the `de1` instead of the `de` request.

2021-06-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/refer-me.tmac:
	* tmac/refer.tmac: Add include guards.

2021-06-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/refer-me.tmac: Handle being loaded in compatibility mode.

2021-06-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Properly situate devtag flag-related
	register names in the package name space.  Rename `need_eo_h` to
	`an-devtag-needs-end-of-heading'.  Rename `need_col2` to
	`an-devtag-needs-second-column`.
	(an-input-trap, SH, SS, HP, (initialization)): Do it.
	(an-input-trap): Test their values more idiomatically.

	Fixes one third of <https://savannah.gnu.org/bugs/?42515>.

2021-06-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (do_if_request): Improve diagnostic
	message; say "conditional operator", not just "conditional".

2021-06-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-end, an-header): In continuous rendering
	mode, use same spacing amount after header and before footer as
	mdoc in its own continuous rendering mode.  Thanks to Ingo
	Schwarze for the suggestion.

	Fixes <https://savannah.gnu.org/bugs/?60653>.

2021-06-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (P): Define this as the "canonical"
	paragraphing macro.
	(LP, PP): Make these aliases of P.

2021-06-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (spring_trap): Tighten lexical
	discipline; describe macro as "trap-called" rather than
	"trap-invoked".

2021-06-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Don't spuriously break a page after a paragraph tag when
	the tag is short enough to fit within the paragraph's
	indentation.

	* tmac/an-old.tmac (an-write-paragraph-tag): Reserve ("need") 2
	vees plus 1 basic unit of vertical space regardless of the width
	of the tag.  In the short tag case, we reverse space by one vee
	after outputting the diversion in which the tag is stored, but
	it will always contain a break, and by the time we reverse, we
	might have already hit a page location trap
	{an-break-body-text}, and by then it's too late to back up.

	* tmac/tests/an-old_no-break-after-short-paragraph-tags.sh: Test
	it.
	* tmac/tmac.am (tmac_TESTS): Run test.

	Fixes <https://savannah.gnu.org/bugs/index.php?60624>.

2021-06-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Rename environments to be more descriptive.
	  an-env -> an-env-header-or-footer
	  an-2   -> an-env-paragraph-tag
	(an-header, an-footer, an-write-paragraph-tag): Do it.

2021-06-14  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac: Add diagnostic error if .EQ used within .TS/.TE.
	Introduce register `tbl@within-table` to track whether we're
	inside a table.
	(@EQ): Check register and issue error if it is true.
	(@TS): Set register.
	(@TE): Clear register.

	Prompted by Savannah #55754.

2021-06-14  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac: Permit document to start with .PS.
	(cov*ab-init): Alias PS to @PS.
	(PS): Call LP, then re-execute (aliased) self as TS does.
	(TS): Update comment to parallel the one in PS.

2021-06-11  Tadziu Hoffmann <hoffmann@usm.uni-muenchen.de>

	Indent an IP paragraph based on the type size of the paragraph
	text, not a preceding heading.  We have to delay evaluating IP's
	argument until the font size has been reset, which I've
	attempted to do by simply passing the argument unevaluated down
	the call hierarchy, instead of saving it in a number register
	right away, as appears customary in this ms implementation.

	However, par@reset is called from a great number of places, and
	these usually set the indent registers beforehand.  I've left
	this as it is, so now two ways of setting the indents exist: via
	register and optionally via argument.

	* tmac/s.tmac (par@reset): Move font-size setting stuff before
	indent-setting stuff; set registers from optional arguments with
	default "n" scaling indicator before setting indents.
	(par*start, par@finish): Don't set registers, but instead pass
	arguments to par@reset.
	(@IP): Pass argument unevaluated to par*start.

	Fixes <https://savannah.gnu.org/bugs/index.php?59604>.

2021-06-11  Tadziu Hoffmann <hoffmann@usm.uni-muenchen.de>

	* tmac/s.tmac (@QP, @XP): Be more scrupulous about using "u"
	scaling indicator.

2021-06-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #59604.

	* tmac/tests/s_IP-indents-using-paragraph-type-size.sh: Add it.
	* tmac/tmac.am (tmac_TEST): Run it.

2021-06-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	Refactor grog even more aggressively.  Make the tedious search
	for subs.pl unnecessary by inlining it into grog.pl directly.
	This makes the script stand alone and much more convenient to
	work with, and doesn't contribute to bloat, in my
	opinion--subs.pl was 10 times the size of grog.pl, and most of
	the code in the latter was dedicated to trying to locate
	subs.pl.

	* Makefile.am: Delete references to $(grog_dir).
	* configure.ac: Delete call of `GROFF_GROGDIR_DEFAULT` macro.
	* m4/groff.m4: Delete definition of same.
	* src/roff/grog/grog.am (GROG, dist_grog_DATA, grogdir): Drop
	variables.
	(grog): Remove target's dependency on `$(GROG)`.  Remove sed
	expression replacing `@grog_dir@` with `$(grog_dir)` in grog.pl.
	* src/roff/grog/grog.pl: Inline most of subs.pl (except for
	author/license comment banner).  Delete sub-hunting logic.
	* src/roff/grog/subs.pl: Delete.

2021-06-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/pic.ms (Basic PIC Concepts, Decorating Objects/Filled
	Objects, PIC Reference/Semi-Formal Grammar): Fix error: the
	keyword "solid" sets a line style, not a fill for a closed
	figure.  Thanks to Dave Kemper and Wim Stockman for reporting
	and researching this issue.

	Fixes <https://savannah.gnu.org/bugs/index.php?60470>.

2021-06-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	* NEWS:
	* doc/webpage.ms:
	* man/groff.7.man (Registers/Read-only registers) <.U>:
	* man/groff_diff.7.man (Language/New number registers) <.U>: Fix
	error; the sense of this Boolean variable was backwardly
	documented {perhaps due to an implementation detail from 2004
	that was reversed in commit 7a0e2f15, 26 February 2008}.

2021-06-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/groff.cpp: Revise usage message.
	(synopsis): Set primary synopsis on a single output line, for
	convenience of parsing and also because we have no control over
	the width of the program name in character cells; it comes from
	argv[0] and is under user control.  There is thus no point in
	trying to get pretty formatting.  Also include separate synopsis
	lines for short and long help options.  Use opposite case from
	option flags for option arguments.
	(help): Collate option flags in English lexicographic order.
	Include space after flag letters for options that take
	arguments.  Consistently use capitals for option arguments
	because they are space-separated here.  Illustrate longer forms
	of -d and -r options.  Update language (some, like "tmac.name",
	was very old).  Add reference to groff(1) man page at end.
	(usage): Drop appendnment of notice regarding -h option; this is
	now covered in `synopsis()` unconditionally.

2021-06-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	Eliminate "groff_opts_no_arg.txt" and "groff_opts_with_arg.txt"
	files from distribution.  They are not needed by any live code
	in the tree; possibly only groffer ever used them.

	* src/roff/groff/groff.am (GROFF_OPTS_OUTPUT, groffoptsdir,
	groffopts_DATA): Delete variables.
	(MOSTLYCLEANFILES): Stop adding to target.
	(groff_opts.tmp, groff_opts_no_arg.txt,
	groff_opts_with_arg.txt): Delete targets.

2021-06-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/grog/subs.pl (version): Report version information in
	a format consistent with our other programs.

2021-06-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/grog/grog.pl: Refactor initialization so that the
	program can be run from a build tree--and tested.  Drop much
	cruft from before Bertrand Garrigues's 2018 work to adopt
	`git-version-gen`.

2021-06-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tests/an-old_title-abbreviation-works.sh: Log more
	accurate messages in test failures.

2021-06-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-abbreviate-title): Add new internal macro
	to reduce the length of a man page title (i.e., the "ls" in
	"ls(1)", if it is too wide to fit in the page header (and would
	overwrite the center header material, usually the title of the
	manual section).  Since the beginning and end of the title might
	be important for disambiguation, truncation at either end would
	be unwise; replace the middle of the title with an ellipsis.
	Define new an-pageref and an-title-abbv strings for use by other
	macros and user-defined PT and BT traps.
	(PT, an-footer): Use new an-pageref string for economy.
	(an-header): Call an-abbreviate-title after setting the title
	length.

	* tmac/tests/an-old_title-abbreviation-works.sh: Test it.
	* tmac/tmac.am (tmac_TESTS): Run test.

	Fixes <https://savannah.gnu.org/bugs/index.php?43532>.

2021-06-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff.7.man (Copy mode): Add new section.

2021-06-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (else_request, while_request,
	pipe_source, open_request, opena_request, pipe_output,
	system_request): Update diagnostic messages to not presume the
	identities of the control or escape characters.

2021-06-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Conventions Used in This Manual): Add node.

2021-05-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/groff.1.man: Expand to aid beginners.
	(initialization): Define `TeX` string.
	(Usage): Rewrite.
	(Usage/Getting started): Add "Hello, world!" example using
	multiple output devices (formats).
	(Usage/Using groff as a REPL): Add example of likely interest to
	programmers.  Or so I hope.  It illustrates the Turing
	fundamentals; we read, store, test, and branch.

2021-05-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/papersize.tmac: Improve diagnostic message; when
	complaining about unrecognized input, report back the input we
	actually received, not what it looks like after we transform it.
	Also add name of macro package to diagnostic so the user knows
	who is speaking, continuing the long process of fixing Savannah
	#52463.

2021-05-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_ms.7.man (Usage/Paragraphs) <QP>: Fix error: .QP
	paragraphs are indented by \n[QI], not \n[PI].

2021-05-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libdriver/input.cpp (do_file): Make fatal diagnostic
	from output driver when the device description file "DESC" can't
	be found more informative by including the name of the device
	for which the input was prepared.  Also makes malformed input of
	intermediate format more obvious.

2021-05-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/tbl.1.man (Miscellaneous): Add paragraph
	cautioning users against making tbl(1) tables conditional
	document content.

2021-05-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Add style warnings for documents with input lines that
	are blank or have leading spaces, since they can cause
	surprising rendering; enabled if the CHECKSTYLE register is
	greater than 2.

	* tmac/an-old.tmac (an-blank-line-trap, an-leading-space-trap):
	Issue style warning if enabled (and only if fill mode enabled in
	leading space trap); in any event, recreate the default *roff
	behavior.
	(TH): Install traps.

	* doc/andoc.tmac (reload-doc): Remove man(7) traps before
	loading doc.tmac.

2021-05-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/andoc.tmac (reload-doc): Fix spurious blank line.

2021-05-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Operators in Conditionals):
	* man/groff.7.man (Control structures/Conditional expressions):
	Fix omission; the conditional operators "F", "m", and "S" also
	accept optional spaces and/or tabs between themselves and their
	arguments.

2021-05-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff.7.man (Requests/Request short reference) <.ta>:
	Describe the request as tersely as possible without omitting its
	essential syntax and semantics.  (The AT&T troff "+" relative
	prefix for normally absolute positions ".ta 1i +1i +1i" and
	alignment-specifying suffixes "L", "R", and "C" had both been
	omitted.)

2021-05-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff.7.man (Registers/Writable registers): Fix error:
	\n[nl] interpolates the current vertical position, not the
	vertical position of the "last printed text baseline"; the
	latter is \n[.h].

2021-05-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Manipulating Filling and Adjustment): Update
	inter-sentence space example to be more illustrative.  Thanks to
	Dave Kemper and Doug McIlroy for the discussion.

	Fixes <https://savannah.gnu.org/bugs/index.php?60404>.

2021-05-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/mdoc/doc-common-u: Add comment warning source divers that
	\n[doc-header-space] is re-used for a purpose for which it is
	not documented.
	* tmac/mdoc/doc-nroff-u: Initialize doc-{header,footer}-space
	registers, which define the vertical spacing between body text
	and the relevant titles (not between the titles and the page
	margins) to 1v instead of 0.5i.  This is what mdoc(7) has always
	done in practice, per Ingo Schwarze.

2021-05-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/devices/grohtml/post-html.cpp (assert_state::add): Avoid
	potential deallocation of statically-allocated strings.  Use
	strsave() to duplicate them so that they can be safely handed to
	a_delete().  Also update diagnostic message to report name of
	complaining program (continuing the long process of fixing
	Savannah #52463).

	Fixes <https://savannah.gnu.org/bugs/index.php?55693>.  Thanks
	to Petru-Florin Mihancea for the report.

2021-05-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/troffrc-end: Fix transposition error (in commit a248aa33
	on 15 January) that broke most table image generation in HTML
	output.

	Fixes <https://savannah.gnu.org/bugs/index.php?60656>.

2021-05-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-ext.tmac (UE, ME): Replace calls to HTML-NS macro for
	emissions of HTML 'A' tags with direct usage of \X device
	control escapes.  Replace .nop requests with text lines since
	this file is supposed to be maximally portable to other *roffs.

2021-05-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/html/pre-html.cpp (main): Call `scanArguments()`
	earlier, before trying to load a font description, so that -v,
	--version, and --help work.

2021-05-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	[grohtml]: Add -C and -G options to postprocessor, to suppress
	output of CreationDate and Creator HTML comments, respectively.
	These can inject unwanted noise into build artifacts.

	* src/devices/grohtml/post-html.cpp: Add static globals
	`do_write_creator_comment` and `do_write_date_comment`.
	(html_printer::{do_file_components,~html_printer}): Write
	comments per corresponding global variables.
	(main): Add to `getopt_long()` call parameter and set as needed.
	(usage): Document them.

	* src/preproc/html/pre-html.cpp (scanArguments): Add to
	`getopt_long()` parameter, but ignore.

	* src/devices/grohtml/grohtml.1.man (Synopsis, Options):
	Document them.

	* src/roff/groff/tests/output_driver_C_and_G_options_work.sh:
	Test them.
	* tmac/tmac.am (groff_TESTS): Run test.

	Fixes 1/3rd of <https://savannah.gnu.org/bugs/index.php?60655>.

2021-05-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/html/pre-html.cpp (usage):
	* src/devices/grohtml/post-html.cpp (usage): Fix inaccuracies in
	supported option summary.

2021-05-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/html/pre-html.cpp (scanArguments):
	* src/devices/grohtml/post-html.cpp (main): Use libgroff's
	warning() instead of printf() to emit diagnostic about
	unrecognized paramter to `-x` option.  This sends the diagnostic
	to the standard error stream and continues the long process of
	fixing Savannah #52463.

2021-05-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_www.7.man: Stop using the "www" macro package.  The
	macros it exposes are not part of the quasi-standardized man(7)
	language, and thus should not be used in man(7) documents.  See
	groff_man_style(7) for portability advice.  Besides, the
	extension macros in the permissively-licensed an-ext.tmac are
	sufficient to support URL and email hyperlinks, and our SH and
	SS macro implementations already support anchor placement
	without requiring explicit effort from page authors.
	(initialization): Stop sourcing www.tmac.  Stop calling .LK and
	.HR; the page-internal navigation links were being placed after
	the apropos line instead of at the beginning of the page,
	contradicting the comment that was there (possibly a bug).
	(Requests) <URL, LK>: Stop calling .TAG.
	<LK>: Drop paragraph advising reader to format page for groff's
	HTML output device to observe the navigation links at that point
	in the test.

2021-05-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Comments): Fix error in description of parser
	operation; '.\"' is an effective whole-line comment not because
	it is a call of an undefined macro named '\"' (no 'mac' warning
	is emitted when that warning type is enabled), but because the
	comment escape is recognized normally and what remains is
	handled as the empty request.  Clarify that the optional
	argument to the .ig request is the name of a macro (which will
	be called normally _and_ end the ignored region; this is
	analogous to ".de foo bar").

2021-05-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-ext.tmac (UR, UE, MT, ME): Rename environment and
	diversion for clarity; they apply to the link text, not the URL
	{or email address}.  The latter might not visibly appear,
	depending on the output driver.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (TH): Remove extraneous right-brace escape.

	Fixes <https://savannah.gnu.org/bugs/?60651>.  Thanks to Bjarni
	Ingi Gislason for catching it.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Copy Mode): Fix omissions from list of escape
	sequences interpreted even in copy mode.  Add \g, \V, "\.",
	'\"', \#, \a, \e and \E (with caveat), and \t.  Parallels
	groff(7) now.  Add cross reference to "Character Translations"
	node.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff.7.man (Escape Sequences): Fix omission; the escape
	character in an unrecognized escape sequence is not merely
	ignored; the sequence produces a warning diagnostic, albeit of a
	type that is disabled by default.
	(Escape Sequences/Escape short reference): Fix omissions; \# and
	\V are interpreted even in copy mode.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff.7.man (Escape Sequences/Escape short reference): Fix
	errors in descriptions of \d, \r, \u escapes; they move in vees,
	not ems.

	Fixes <https://savannah.gnu.org/bugs/?60602>.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-header): Fix missing brace escape.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Slightly refactor footer spacing management when
	continuously rendering.

	* tmac/an-old.tmac (an-end): Increase the page length _after_
	flushing a pending output line.  Do so only by the amount of
	spacing we actually perform next.  Make the scaling indicator on
	the .sp request explicit.  The amount seemed magical because it
	was accounting for the one line to be consumed by output of the
	footer title line.  Instead, move that further increase of the
	page length from here...
	(an-footer): ...to here (~250 lines away), before calling BT.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/mdoc/doc-common-u (doc-end-macro): When continuously
	rendering and after flushing the last line of the body text of a
	page, vertically space by 3 vees instead of 1, for consistency
	with our man(7) implementation.  Also increase page length by
	same amount _after_ flushing a pending output line, for symmetry
	with other spacing requests (and to prevent nasty surprises
	analogous to those in Savannah #60611).
	(doc-header): Put 3 vees of space after the header in continuous
	rendering mode, not 1 (and increase page length accordingly).

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Add warnings for macros deprecated in groff_man(7),
	emitted if the CHECKSTYLE register is greater than 1.

	* tmac/an-old.tmac (an-deprecation-warn): New macro emits
	diagnostic.
	(DT, PD): Call the foregoing, re-using arguments.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Refactor to move bodies of DT and PD into
	private macros.
	(an-reset-tab-stops, an-reset-paragraph-spacing): New names for
	the former DT and PT.
	(TH): Call these new macro names.
	(DT, PD): Wrap the corresponding private macros.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-bp): In continuous rendering mode, draw
	a horizontal line between a page footer of one man page document
	and the header of the next, but only if multiple documents are
	being rendered.
	(initialization): Use variable to ensure we don't draw this line
	after finishing the only document we render.

	Fixes other half of <https://savannah.gnu.org/bugs/?60611>.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Fix stochastic vertical space loss problems in continuous
	rendering mode by being more scrupulous about extending the page
	length.  As noted in Savannah #60611, sometimes the vertical
	space between a header line and the first section heading
	{"Name"} would be reduced or removed altogether, in a way that
	was highly sensitive to circumstances.  As far as I can tell,
	the problem is that this vertical space disappeared only when we
	were in double-secret probation no-space mode; this is the
	special non-spacing mode {not apparent with inspection of
	\n[.ns]} that happens when the stealth internal page location
	trap is sprung.  This trap does not appear in .ptr output but
	can be disabled with ".vpt 0", which is an alternative fix for
	this problem {because continuous rendering mode does not use
	vertical position traps at all}.  Another fix might have been to
	simply call the moral equivalent of ".pl MAXINT" as soon as
	continuous rendering mode was decided upon.

	* tmac/an-old.tmac (an-header): In continuous rendering mode,
	extend the page length by one line before calling PT to output
	the page header, and afterwards, extend the page length by 3
	vees and space by that amount (instead of half an inch).

	Fixes half of <https://savannah.gnu.org/bugs/?60611>.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-header, an-footer): Drop no-op .tl
	requests.

2021-05-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-footer): Improve symmetry with an-header.
	If rendering to HTML, don't bother setting the
	an-outer-footer-text, an-extra1, and an-extra2 strings to empty
	values; instead don't call BT, just as we don't call PT in
	an-header.

2021-05-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-end): Drop needless .nr % assignment.
	Call our (wrapped) bp instead of the .pl request directly.
	(an-bp): Drop needless .br and (aliased) real .bp request.

2021-05-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-footer): Refactor complex conditional.

2021-05-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tests/an-old_X-register-works.sh: Add test for footer
	creeping into page content.

2021-05-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (initialization): Ignore P and X rendering
	parameter registers if specified with continuous rendering
	enabled (the default in the nroff mode used for terminals), and
	emit diagnostic.

2021-05-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tests/an-old_X-register-works.sh: Rewrite.  Increase test
	coverage and stop assuming that -rcR=1 and -rX=anything is a
	supported combination.  (Continuous rendering is supposed to
	suppress the printing of page numbers anyway.)

2021-05-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (initialization): Refactor handling of P and
	X rendering parameter register validation in preparation for
	disabling them in continuous rendering mode, where they don't
	make any sense.

2021-05-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tests/an-old_TH-repairs-ad-damage.sh: Tweak test
	internals to make troubleshooting easier.

2021-05-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (TH): Relocate some initialization logic to
	be adjacent to (most of) the rest of it.

2021-05-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (TH): Remove useless .if.  Its condition is
	always true now in the wake of the fix for Savannah #60609.

2021-05-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Rename many package macros, strings, and
	registers (and the lone named environment) to be more expressive
	of their function.  "an-footer" is now the proper complement of
	"an-header" (setting up an environment and calling a
	user-redefinable trap macro).
	  an-first         -> an-is-first-page-of-document
	  an-html          -> an-is-output-html
	  an-footer        -> an-break-body-text
	  an-p-footer      -> an-footer
	  an-set-margin    -> an-reset-margin-and-inset-level
	  an-level         -> an-inset-level
	  an-tag-sep       -> an-tag-separation
	  an-no-space-flag -> an-need-no-space-mode
	  an-break-flag    -> an-need-break
	  an-div?          -> an-is-in-diversion
	  an-page-string   -> an-outer-footer-text
	  an-trap          -> an-input-trap
	  an-do-tag        -> an-write-paragraph-tag
	  an-1             -> an-env

	* tmac/andoc.tmac (reload-doc): ...as above for the exposed
	parts of the interface.
	  an-footer        -> an-break-body-text
	  an-p-footer      -> an-footer

2021-05-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Rearrange alternating font macros to be in
	alphabetical order (and the order documented in groff_man(7).

2021-05-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Refactor alternating font macros.  These
	macros are hard enough to understand without adding gratuitous
	structural differences to the reader's burden.  Make their
	implementations rigidly parallel.  Update comments.
	(RI, IR, IB, BI, RB, BR): Always define the `an-result` string
	as empty except for a non-printing input break for the sake of
	compatibility mode.
	(RI, IR, IB, BI): Defer interpolation of the first argument to
	the while loop if there are at least two (like the existing RB,
	BR).  This way the the font escape style names in the appendment
	to `an-result` match the order of the letters in the macro name
	{useful to make sense of diffs}.
	(RI, IR, IB, BI, RB, BR): Always remove the `an-result` string
	after interpolating it.
	(RB, BR): Only do work if there are any arguments.

2021-05-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Apply italic corrections more consistently.

	* tmac/an-old.tmac (BI, IB, IR, RI): Apply italic corrections
	more consistently.  Add lengthy comment rationalizing approach.
	(RI, BI): An italic correction is now applied after the last
	argument if it is of even parity (i.e., the 2nd, 4th, 6th, ...).

	* tmac/groff_man.7.man.in (Description/Font style macros): Note
	that italic corrections are applied.

	Fixes <https://savannah.gnu.org/bugs/index.php?53541>.

	Thanks to Bjarni Ingi Gislason for the report and a suggested
	patch.

2021-05-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Ignore FT register when appropriate.

	* tmac/an-old.tmac (initialization): Validate user-specified
	value of \n[FT].  Continuous rendering ignores FT (page location
	traps are not used for footers in that mode).  Measuring a
	footer distance from the page top isn't done.  A footer distance
	of over half the page length is unlikely.  A footer distance of
	less than one line height is too.  Issue diagnostics in all of
	these cases.

	* tmac/tests/tmac/tests/\
	an-old_FT-bad-value-should-not-trash-titles.sh: Test it.
	* tmac/tmac.am (tmac_TESTS): Run test.

	Fixes <https://savannah.gnu.org/bugs/index.php?60612>.

2021-05-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (TH): Plant the an-footer trap (where we
	schedule a break in the page text prior to spacing down to the
	footer) at twice the value of \n[FT], not a hard-coded -1i.
	This way, the user can't make the an-p-footer trap (planted at
	\n[FT]) spring before the an-footer trap that should precede it.

	* tmac/groff_man.7.man.in (Options) <-rFT>: Document this.

2021-05-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* m4/groff.m4 (GROFF_GHOSTSCRIPT_VERSION_CHECK): Add new
	autoconf macro to check Ghostscript version.  Versions 9.00 <= x
	< 9.54 suffer from a rendering glitch that affects the AT&T
	troff (and groff) special character \(lh; see
	<https://bugs.ghostscript.com/show_bug.cgi?id=703187>.  Store
	the result of the check.
	(GROFF_GHOSTSCRIPT_VERSION_NOTICE): Add new macro to report the
	problem detected by GROFF_GHOSTSCRIPT_VERSION_CHECK to the user.

	* configure.ac: Call the macros in appropriate places.

	Fixes <https://savannah.gnu.org/bugs/index.php?59531>.

2021-05-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/andoc.tmac (reload-man): Remove removal of RI macro, made
	unnecessary by commit 551f138 (15 May).

2021-05-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* m4/groff.m4 (GROFF_HTML_PROGRAMS, GROFF_PDFDOC_PROGRAMS):
	Tweak autoconf warning messages when Ghostscript (gs) command
	missing to be more mutually congruent.
	(GROFF_URW_FONTS_CHECK): Tweak to note that gs is only run to
	locate the URW fonts if gs is available.

2021-05-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tests/an-old_AT-and-UC-footer-saved-and-restored.sh:
	Tweak to be more helpful when troubleshooting exasperating
	header/footer problems as in Savannah #60609.

2021-05-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Fix missing page footers when continuously rendering
	multiple documents.  Resolving this uncovered entangled header
	and footer management bugs with batch processing even when _not_
	continuously rendering.

	* tmac/an-old.tmac (an-start-new-document): New macro clears
	header trap so it doesn't get called with stale information.
	{The TH macro already replants a header trap with every man
	page if not continuously rendering.}  This macro also absorbs
	the break-page and page number incrementation logic from TH.
	(TH): Adapt semantics of "an-first" register to mean not "first
	page rendered" (apparently) but "first page of new document".
	If we are _not_ on the first page of a new document and the TH
	macro is called, we must be batch processing: force the previous
	man page to end (call an-end if continuously rendering, or
	an-start-new-document if not).  After all the header and footer
	strings have been configured for the man page, call an-header if
	we're on the first page of the document, not (just) if we're
	continuously rendering.  Zero an-first register at end of macro
	here unconditionally instead of conditionally in .an-header.
	(an-header): Drop conditional on continuous rendering.

	When I batch-render our 62 man pages to the terminal, I do note
	one cosmetic regression: the lj4_font(5) page, when continuously
	rendered in batch processing, has no space between its header
	line and first section heading ("Name").  Another cosmetic issue
	in this scenario is that the footers for every rendered page but
	the last, which had (wrongly) been absent entirely, are set with
	no space after them and the header of the next.

	* tmac/tests/an-old_page-footers-present.sh:
	* tmac/tests/an-old_page-header-has-current-data.sh: Add man
	page header/footer regression tests.
	* tmac/tmac.am (tmac_TESTS): Run the tests.

	Fixes <https://savannah.gnu.org/bugs/index.php?60609>.

2021-05-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (initialization): Key the reprocessing guard
	{like a C language #include guard} on the existence of the TH
	macro instead of RI.  The latter seemed kind of random and TH
	is even more guaranteed to exist; it is the macro name most
	characteristic of the man(7) language and the andoc.tmac
	superstructure relies on it.  (TH may have been avoided because
	of its use with tbl(1); however, neither man(7) nor mdoc(7)
	support tbl's TH feature, and GNU tbl does not supply its own
	empty fallback definition as it does for TS and TE.  If we ever
	want TH-like functionality in future man pages, I suggest we
	pick a new, non-colliding name for it.)
	(initialization): Relocate load of devtag package; we don't
	need it before entering compatibility mode.  Move assignment of
	format of an-page-letter register from a "loose declaration"
	amid macro definitions to within existence test of X register.
	Stop initializing an-extra[123] strings outside of any macro;
	they need to be reinitialized at every TH call when batch
	rendering.
	(TH): Call DT in a less surprising place.
	(TH): Initialize all header/footer-related strings before
	calling header macros, even those not needed for the default
	header configuration.
	(TH): Add style warnings for underspecified arguments.
	(an-ne): Rename register from "an-need", which is pretty
	confusing in context, to "an-amount", since (1) a (vertical)
	amount is what it is, and (2) it's only used within this macro.
	Also due to factor (2), remove the register after using it.

2021-05-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	* m4/groff.m4 (GROFF_URW_FONTS): Update configure script check
	for URW fonts to recognize recent releases of the URW Base 35
	fonts from Artifex Software.  Look for the AR (Arial roman
	equivalent) font under all of the names recognized by
	fonts/devpdf/Foundry.in.
	* m4/groff.m4 (GROFF_URW_FONTS_CHECK): Update wording of notice
	when URW fonts are not found.  Replace dead URL with working
	one.  Try to hedge against the future, but the Web is an
	ephemeral place, as are the file names Artifex gives to its
	fonts.  Thanks to T. Kurt Bond and Thomas Dupond for reporting
	and investigating the problem.

	Fixes <https://savannah.gnu.org/bugs/index.php?60604>.

2021-05-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (ms Document Control Settings) <DI>:
	* doc/ms.ms (Document control settings, Displays and keeps)
	<DI>:
	* tmac/groff_ms.7.man (Document control settings, Displays and
	keeps) <DI>: Document this register; it had been mentioned only
	in passing.

2021-05-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (ms Document Control Settings) <DD>:
	* doc/ms.ms (Document control settings) <DD>:
	* tmac/groff_ms.7.man (Document control settings) <DD>: Fix
	error; changes to \n[DD] take effect at the next display
	boundary (including the end of the active one), not the next
	"paragraph" (paragraph macros cannot be called inside a display
	anyway).

2021-05-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/andoc.tmac (reload-man): Delete no-op line that attempts
	to restore compatibility mode.  Individual pages have to do
	this, and ".do cp \n(.C" is not effective for the reason
	discussed in the 2020-04-16 ChangeLog entry.

2021-05-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac (@RT): Delete definition; make it an alias for
	par@finish.  Until commit 021ba0e7 (1 May), they had identical
	definitions.  I find it difficult to imagine what use it would
	be to have an undocumented reset macro that resets everything
	_except_ the deeply internal \n[.ev]:ai.

2021-05-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: Handle HF strings with an embedded font family when
	applying italic-to-bold-italic remapping; the feature now works
	on troff devices as well.

	* tmac/an-old.tmac (initialization): Do more validity checking.
	Store heading family in new string \*[an-heading-family].  Clean
	up after self.
	(SH, SS): Include the heading family in the remapping target.

	* tmac/groff_man.7.man.in (Description/Document structure
	macros) <.SH, .SS>: Document it.

2021-05-07  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff_char.7.man (Glyph tables/Mathematical symbols): Fix
	errors in special character descriptions; \[sqrt] is a "special"
	glyph (that is, it uses mathematical metrics on typesetter
	devices), and \[radicalex) is not (and is thus a "text" glyph).
	See Werner Lemberg's ChangeLog entry of 2003-01-05.  Problem
	dates to 07a6233ad, 27 May 2014.

	The overloaded use of the word "special" really pinches here.

2021-05-07  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tty.tmac: Replace fallback glyphs for radical extension
	and square root extension; use \[rn] instead, which works fine.

	Now the only glyph in the page that doesn't render on a UTF-8
	terminal (if one's font has adequate coverage) is the Bell
	Systems logo.

2021-05-07  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tty.char: Add fallback characters for \[fm] and \[sd].

2021-05-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	[man]: If \*[HF] is a bold style, substitute bold italics for
	italics in section and subsection headings, keeping the font
	weight consistent.

	* tmac/an-old.tmac (initialization): Set a flag for this based
	on the interpolation of \*[HF] matching "B" as its last
	character.
	(SH, SS): Apply and reverse font remapping based on this flag.

2021-05-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-ext.tmac (EX): Work around Savannah #59522 by changing
	fonts differently on DVI output, avoiding a font warning.

2021-05-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-ext.tmac: If in nroff mode, remap font CBI to BI.
	* tmac/dvi.tmac: Remap font CBI to CWI.  (Computer Modern
	Constant Width has no bold styles.)

2021-05-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac (NL, SM, LG): Issue warning diagnostic if macro
	called with arguments.
	(par*define-font-macro): Construct font macro with diagnostic to
	warn if called with excess arguments.
	(UL, BX): Issue warning diagnostic if macro called with excess
	arguments.

	Fixes <https://savannah.gnu.org/bugs/index.php?60522>.

2021-05-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi: Delete redefinition of \putwordAppendix
	Texinfo macro.  This restores the word "Appendix" to the names
	of all appendices and fixes two bad internal links to Appendix
	E, the Register Index, in the DVI and PDF output formats.  Per
	consultation with Texinfo maintainer Gavin Smith, it's difficult
	to get the behavior we want, so give up trying for now.

2021-05-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac (@MC): Actually diagnose and recover when user
	tries to .MC inside a diversion.

2021-05-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Differences from AT&T ms):
	* doc/ms.ms (Differences from AT&T ms):
	* tmac/groff_ms.7.man (Differences from AT&T ms): Document our
	different default (empty) for the center footer in nroff mode;
	this behavior appears to date back to June 1991 or earlier,
	drawing few complaints.

	Fixes <https://savannah.gnu.org/bugs/?59826>.

2021-05-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tests/*: Rename files to use consistent scheme; one
	underscore to separate the package name from the test objective,
	dashes to otherwise separate words.
	* tmac/tmac.am (tmac_TESTS): Use new names.

2021-05-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	[ms]: Restore support for AT&T ms PN register.

	* tmac/s.tmac (pg): Make register PN an alias of %.

	* doc/groff.texi (Differences from AT&T ms):
	* doc/ms.ms (Differences from AT&T ms):
	* tmac/groff_ms.7.man (Differences from troff ms): We don't need
	PN, and it was removed in commit 08291b40 (25 October 2020), but
	since it is documented in Lesk 1978 we're stuck with it.  Advise
	users to stick with %.  Also warn them of the hoop they must
	jump through if they redefine the page trap macro PT, which Lesk
	1978 also encourages.

	* tmac/tests/s_PN-works.sh: Test it.
	* tmac/tmac.am (tmac_TESTS): Run test.

	Fixes <https://savannah.gnu.org/bugs/?59823>.

2021-05-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	[ms]: Restore default IP paragraph indentation after an SH call.

	* tmac/s.tmac (par@finish): Set "ai" register for the current
	environment to the paragraph indent (PI) setting for the current
	environment.

	* doc/groff.texi (Highlighting in ms) <IP>:
	* doc/ms.ms (Highlighting) <IP>:
	* tmac/groff_ms.7.man (Usage/Highlighting) <IP>: Update
	documentation.

	* tmac/tests/s_SH_resets_IP_indentation_amount: Test it.
	* tmac/tmac.am (tmac_TESTS): Run test.

	Fixes <https://savannah.gnu.org/bugs/?60222>.

2021-05-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac (DS): Issue diagnostic and attempt recovery if
	document attempts to start a display within another display.

	Fixes <https://savannah.gnu.org/bugs/?55136>.

2021-04-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac: Implement alternative rendering for automatic
	footnote numbers in nroff mode, inspired by Heirloom Doctools
	ms; since superscripting is not univerally available on
	terminals, surround number with square brackets instead.
	(fn@init): New macro finishes initializing footnote module,
	setting up aliases to par@sup-{start,end} strings in troff mode
	and otherwise defining fn@sup-{start,end} as brackets.  Define *
	string here, using new strings.
	(par): Call fn@init after par@sup-{start,end} are defined, and
	prior to other footnote integration material.
	(par*fp!0): Migrate from par@sup-{start,end} to
	fn@sup-{start,end}.

	* doc/groff.texi (ms Document Control Settings) <FF>:
	* doc/ms.ms (Footnotes) <FF>:
	* tmac/groff_ms.7.man (Usage/Footnotes) <FF>: Document it.

	Fixes <https://savannah.gnu.org/bugs/?60228>.

2021-04-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* test-groff.in: Stop passing the built groff executable -b and
	-ww options by default.  This reverses a change from 2017.  Dave
	Kemper has convincingly argued that there's no way to override
	-b and turn backtraces on warnings/errors back off, and that the
	built groff should behave as much like a "normal" one as
	possible.  The latter factor is of added importance now that we
	are using test-groff for generation of documents.

	* src/roff/groff/tests/regression_savannah_58153.sh: Adapt.

2021-04-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac (BX): Add alternative implementation for terminal
	{nroff} devices.  Store width of boxed text, adding .4m only on
	troff devices (to make room for the vertical box lines).  Break
	long input line in troff implementation.  If not in troff mode,
	use ISO 6429 color escapes to render boxed text in black on
	white.  Use the \Z escape to match breaking semantics of macro
	in troff mode.

	* doc/groff.texi (Highlighting in ms) <BX>:
	* doc/ms.ms (Highlighting) <BX>:
	* tmac/groff_ms.7.man (Usage/Highlighting) <BX>: Document it.

	Fixes <https://savannah.gnu.org/bugs/?60477>.

2021-04-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/doc.am: Make $(PROCESSEDDOCFILES) depend on all the same
	targets as $(PROCESSEDEXAMPLEFILES); this way the me, ms, and
	pic.ms manuals are all regenerated upon changes to their
	underlying macro packages.

2021-04-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* INSTALL.extra: Update.
	(Evaluation): New section updates material on "test-groff"
	script.  Also tell people how to preview our Texinfo manual (in
	info, text, DVI, PDF, and HTML formats).
	(In Case of Trouble): New section updates bug reporting URL.

	Fixes <https://savannah.gnu.org/bugs/?57863>.

2021-04-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac: Make the FAM string work more sensibly; it now
	applies to headers, footers, and footnotes only if set
	before the first call of a sectioning, paragraphing, or
	{non-date} document description macro.
	(PT, BT): Set the font family to that saved for titles.
	(fn*do-FS): Set the font family to that saved for footnotes.
	(par@init): Copy the document font family to independent strings
	for titles and footnotes.
	(par@reset): If in a footnote environment, set the font family
	to that saved for footnotes; otherwise use \*[FAM].

	* tmac/groff_ms.7.man (Differences from troff ms/Text settings):
	Describe placement-dependent effect of FAM string setting.

	* doc/groff.texi (Highlighting in ms):
	* doc/ms.ms (Highlighting): As above, and suggest different
	occasions in which ".ds FAM C" and "CW" are convenient.

	Fixes <https://savannah.gnu.org/bugs/?60422>.

2021-04-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (source, source_quietly,
	macro_source, macro_source_quietly): Use idiomatic Boolean
	literals.

2021-04-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi: Drop workarounds for Texinfo pre-5.0 versions;
	our local macros produced many warnings during generation of our
	Texinfo manual.
	(Langlemacro):
	(Ranglemacro):
	(Lparenmacro):
	(Rparenmacro):
	(Lbrackmacro):
	(Rbrackmacro): Delete.  Replace call sites with literals.
	(Lbracemacro):
	(Rbracemacro): Delete.  Replace call sites with @lbracechar{}
	and @rbracechar{}.
	(angles): Reimplement in terms of @guilsinglleft{} and
	@guilsinglright{}.  Flesh out comment.

	Fixes remainder of <https://savannah.gnu.org/bugs/?59524>.

2021-04-17  Dave Kemper <saint.snit@gmail.com>

	* doc/meintro.me:
	* doc/meref.me: Correct and make consistent usage of the term
	"point size".

	Fixes <https://savannah.gnu.org/bugs/?60403>.

2021-04-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (SS): Add devtags support.  Set a subsection
	title as a second-level heading and add it to a table of
	contents, if any.  Prompted by a query from Hans Unzner to the
	groff mailing list.  Thanks, Hans!

2021-04-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (SH): Fix apparent thinko.  Pass the devtags
	macro a literal "1" argument.  Passing it \n[an-level] wasn't
	correct (it stores the relative inset level, not a sectioning
	level), and was useless anyway: SH has already called
	.an-set-margin by this point, which resets \n[an-level] to 1.
	man(7) does not support nesting of SH macros.  Also call
	.DEVTAG-NH by its alias .DEVTAG-SH since section headings are
	not numbered in man(7).

2021-04-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	[tmac]: Adjust editor file encoding hints.

	* tmac/cs.tmac: Tell Vim the file is "iso-8859-2" (ISO Latin-2).
	* tmac/en.tmac: Stop telling Emacs the file is "latin-1"; it's
	ASCII.
	* tmac/fr.tmac: Tell Vim file is "iso-8859-15" (ISO Latin-9).

	Vim users may need to use ":e ++enc=iso-8859-2", for instance,
	to see correct glyphs.

2021-04-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	The first-order determinant of hyphenation points is language,
	not territory.  Use ISO 639 2-letter language codes for
	hyphenation and exception patterns instead of ISO 3166 2-letter
	territory codes.

	* tmac/*.us: Rename *.us files to *.en.
	* tmac/troffrc: Change hyphenation language "us" to "en".
	* tmac/en.tmac: Add English localization file.  Set hyphenation
	mode to 4.
	* tmac/troffrc: Derive groff locale from system.

	* doc/groff.texi (Manipulating Hyphenation):
	* man/groff.7.man (Hyphenation):
	* man/groff_diff.7.man (Implementation differences): Refer to
	"U.S. English" hyphenation patterns as simply "English"; they
	will be mostly correct for Commonwealth English as well, and no
	alternative English hyphenation patterns for other territories
	are available.

	* doc/groff.texi (Manipulating Hyphenation):
	* man/groff_diff.7.man (New requests): Note that default
	hyphenation mode depends on the language used on the system.
	Add concept index entry for localization.  Add file index
	entries for the locale macro files (cs.tmac, etc.).  Update to
	reflect rename of English hyphenation patterns and .hla
	identifier from "us" to "en".

	* src/roff/groff/tests/localization_works.sh: Add 10 test cases.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

	* tmac/LOCALIZATION: Rewrite.

	Fixes <https://savannah.gnu.org/bugs/?59814>.

2021-04-12  Dave Kemper <saint.snit@gmail.com>

	* doc/meref.me: Correct various small issues.

	Fixes <https://savannah.gnu.org/bugs/?60379>.

2021-04-12  Dave Kemper <saint.snit@gmail.com>

	* man/groff.7.man (Registers/Writable registers) <year, yr>:
	Drop discussion of Y2K issues.

	Fixes <https://savannah.gnu.org/bugs/?60372>.

2021-04-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Setting Registers) <rnn>: Fix error: the
	request is not ignored if the second argument (the new name)
	does not exist.  Problem dates back to 52a6d12a (11 May 2000).

2021-04-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	Implement new .soquiet and .msoquiet requests.

	Needed for two planned developments: {1} the i18n patch in
	Savannah #59814 can use this to quietly attempt to open a
	localization file.  groff only supports a few locales, so people
	with LANG=es_ES, for instance, would ordinarily see warning
	diagnostics on every groff startup; and {2} system- or
	site-configurable support of per-user {.,}troffrc or man.local
	files, gracefully failing if they don't exist.

	* src/roff/troff/input.cpp:
	(do_source): Add new function, taking boolean "quietly"
	parameter and absorbing function of source(), with added
	conditional.
	(source): Convert into a wrapper to call do_source() unquietly.
	(source_quietly): Wrap do_source(), quietly.
	(do_macro_source, macro_source, macro_source_quietly): Analogous
	to the foregoing.
	(init_requests): Hook "msoquiet" to macro_source_quietly() and
	"soquiet" to "source_quietly().

	* doc/groff.texi (I/O):
	* man/groff.7.man (Requests/Request short reference):
	* man/groff_diff.7.man (Language/New requests): Document them.

	* src/roff/groff/tests/msoquiet_works.sh:
	* src/roff/groff/tests/soquiet_works.sh: Test them.

	* src/roff/groff/groff.am (groff_TESTS): Add tests.

	Fixes <https://savannah.gnu.org/bugs/?59973>.

2021-04-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (ms Document Control Settings) <FF>: Update
	description to cover application of footnote format to footnote
	markers in general, not just numbers.  Clarify different
	behavior of format 1 with respect to automatic numbers and other
	markers.
	(ms Footnotes) <*>: Describe string operation in more detail.
	<FS>: Describe more precisely how the optional argument is
	handled.

	* doc/ms.ms (Footnotes): Synchronize with doc/groff.texi.  Add
	example using document's own text as a model.

	* tmac/groff_ms.7.man (Usage/Footnotes): Resync relevant
	portions with doc/ms.ms.

	Fixes <https://savannah.gnu.org/bugs/?60227>.

2021-04-06  Dave Kemper <saint.snit@gmail.com>

	* src/roff/nroff/nroff.sh: Recognize -k and -K options and pass
	them through to troff.  Document them in usage message.
	* src/roff/nroff/nroff.1.man: Document new -k and -K options.

	Fixes <https://savannah.gnu.org/bugs/?60349>.

2021-04-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Manipulating Hyphenation):
	Recast introductory paragraph to better distinguish the
	automatic breaking of words and placement of hyphens at
	user-specified hyphenation points (done with the \% escape) and
	automatic determination of hyphenation points within words (what
	the pattern files and most of the hyphenation mode parameters
	are for).  Clarify that what our manual calls "automatic
	hyphenation" pertains only to the latter.
	<hw> Correct over-general claim; hyphenation exceptions defined
	with the .hw request _do_ honor .hy value 2 (don't break a word
	at the bottom of a page), but none of the others.
	<\%> Recast description of escape to emphasize independence of
	its two uses (e.g., "\%pseudo\%unununium" is hyphenated only
	after "pseudo-", if at all).
	<\:> Clarify that escape is an input word boundary.
	<shc> Clarify that the soft hyphen glyph is applied to manual as
	well as automatically-determined hyphenation points.
	<hy> Recharacterize as setting the _automatic_ hyphenation mode.
	Note that restrictions apart from value 2 are not applied to
	words with manually-assigned hyphenation points.
	<nh> Note that request affects only automatic hyphenation.

	* man/groff.7.man (Requests/Request short reference) <.hy,.nh>:
	Clarify that requests deal with _automatic_ hyphenation.
	(Registers/Read-only registers) <.hy>:
	Clarify that register applies to _automatic_ hyphenation.
	(Hyphenation): Add new introductory paragraph summarizing manual
	hyphenation support and features, including notice that breaking
	at explicit hyphens is performed in fill mode, i.e., even if
	_automatic_ hyphenation is disabled.

	Thanks to Peter Schaffter for the report.

	Fixes <https://savannah.gnu.org/bugs/?60332>.

2021-03-27  Deri James  <deri@chuzzlewit.myzen.co.uk>

	Changes to the ghostscript fontnames (9.53.3).

	* font/devpdf/Foundry.in: HI and HBI switch to using
	Italic rather than Oblique.

2021-03-24  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Manipulating Filling and Adjustment):
	* man/groff_diff.7.man (Language/Extended requests): Fix
	incorrect claim that additional inter-sentence space is applied
	only in fill mode.  Drop word "parameter" from description of
	unit used in .ss request arguments; it is a needlessly specific
	reference to the font file format.  Tighten wording.

	* man/groff.7.man (Control Characters): Fix incorrect claim that
	end-of-sentence detection is attempted only in fill mode.  Note
	that .ss request also affects inter-word spacing.

	* man/groff_diff.7.man (Language/Extended requests):
	  - Drop example.
	  - Move discussion of AT&T vs. GNU troff difference in .ss
	    handling from here...
	  (Implementation Differences): ...to here.

	Thanks to Dave Kemper for his continued scrutiny!

2021-03-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (ms Cover Page Macros):
	* doc/ms.ms (Cover page macros):
	* tmac/groff_ms.7.man (Usage/Cover page macros): Revise and
	rename (sub)sections to "Document description macros".

2021-03-15  Dave Kemper <saint.snit@gmail.com>

	* Makefile.am: Fix typos.  Thanks to Bjarni Ingi Gislason for
	the report.

	Fixes <https://savannah.gnu.org/bugs/?59457>.

2021-03-14  Dave Kemper <saint.snit@gmail.com>

	* man/groff_tmac.5.man: Correct erroneous reference to PSPIC
	where PDFPIC was intended.  Fix numerous smaller style and
	content problems.

	Fixes <https://savannah.gnu.org/bugs/?60229>.

2021-03-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Optional man extensions) <CW, EX, PN, Pn>:
	(Highlighting in ms) <CW>:
	(Additional ms macros) <CW>: Use "monospaced" to refer to font
	selected, retaining "constant-width" term only where it explains
	the macro mnemonic.  Thanks to Dave Kemper for pointing out the
	issue.  Also fix error: .CW in the ms package seems to have
	originated with Research Unix, not Berkeley.

	* doc/ms.ms (Highlighting) <CW>: Sync with our Texinfo manual.
	* tmac/groff_ms.7.man: Sync with ms.ms, omitting FAM advice.

	Fixes <https://savannah.gnu.org/bugs/?60037>.

2021-03-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Identifiers, Strings, Writing Macros,
	Diversions): Improve cross-linkage of documentation regarding
	shared name space of macros, strings, diversions, and boxes.

	Fixes <https://savannah.gnu.org/bugs/?58995>.

2021-03-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac: Improve diagnostic messages.
	(@diag): New macro wraps .tm request, always prefixing it with
	the name of the macro package, as well as file and line
	indicators only if these are set to non-empty or non-zero
	values, respectively.
	(@error, @warning): Call @diag instead of .tm directly.
	(@fatal): Retire; it had only one call site and it would have
	greatly complicated @diag to support calling .ab instead of .tm.
	An ugly alternative would have been to call .ab with a redundant
	message after calling @diag.
	(@divert): New macro wraps .di request, remembering the name of
	the current file (\n[.F]) when a diversion is started.
	(@error-recover): Call .ab instead of (deleted) @fatal.
	(pg@super-eject): Tell user what the last file name seen was
	when recovering from an unfinished diversion when processing
	ends.

	Fixes <https://savannah.gnu.org/bugs/?55099>.

2021-02-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	Fix bug where having line numbering off but the output line
	number register \n[ln] set to a positive value would cause
	tbl(1) table rows to spontaneously become numbered.  Use new
	\n[.nm] register to determine whether line numbering is enabled.

	* src/preproc/tbl/table.cpp (table::init_output, table::do_row,
	table::do_bottom): Predicate all conditions on \n[ln]
	additionally on \n[.nm].

	Thanks to Olle Lögdahl for the report.  Problem appears to date
	back to commit b69062693d3360efce9d4d63fac337be21e07db7, 20 July
	2011.

	Fixes <https://savannah.gnu.org/bugs/?59812>.

2021-02-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #59812.

	* src/preproc/tbl/tests/cooperate-with-nm-request.sh: Do it.
	* src/preproc/tbl/tbl.am (tbl_TEST): Run test.

2021-02-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	Implement new read-only register ".nm".  It reports the
	enablement status of output line numbering (caused by the .nm
	request) irrespective of the temporary suspension of numbering
	with the .nn request.  Needed because there was no way to
	introspect its state, tbl(1) needs to do so, and the writable
	line number register \n[ln] is not a reliable proxy for it.

	* src/roff/troff/env.h: Add get_numbering_nodes(), returning
	`int`, to public interface of `environment` class.
	* src/roff/troff/env.cpp: Implement get_numbering_nodes().

	* src/roff/groff/tests/dot-nm_register_works.sh: Test it.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

	* doc/groff.texi (Miscellaneous):
	* man/groff_diff.7.man (Language/New number registers):
	* man/groff.7.man (Registers/Read-only registers): Document it.

	* NEWS: Add item.

2021-02-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac: Emit warning diagnostic when using a multi-page
	table with a repeating header (".TS H").
	(KS): Rename diversion from "kp*div" to "kp@div" since the name
	is now referenced outside the keep module.
	(KF): ...similarly for kp@fdiv.
	(generally): Update diversion dereference sites and derived
	names.
	(@TS): When handling "H" argument, check identity of current
	diversion and emit appropriately worded warning.

2021-02-14  G. Branden Robinson <g.branden.robinson@gmail.com>

	* FOR-RELEASE: Add more procedures and information related to
	incrementing groff's version number.

2021-02-14  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libgroff/searchpath.cpp (search_path::open_file,
	search_path::open_file_cautious): Save errno before calling
	free() and restore it afterwards.  A future version of POSIX
	will require that free() not change errno if it succeeds[1];
	some C library implementations, including recent versions of
	glibc[2], lack this property.  free() is called in these
	libgroff functions to clean up after an unsuccessful fopen() of
	a heap-allocated file name string, and because the errno from
	fopen() may be passed to strerror() in a diagnostic message, it
	needs to be accurate.  I checked the rest of groff's codebase
	and found no other instances of free() being used to clean up
	after fopen() failure.

	[1] https://www.austingroupbugs.net/view.php?id=385
	[2] https://sourceware.org/bugzilla/attachment.cgi?id=13073

2021-02-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	Address build failure on macOS.

	Follow the advice of the gnulib manual ("Changing your source
	for use with gnulib") more scrupulously; include config.h in
	more files.

	* src/libs/libgroff/curtime.cpp [HAVE_CONFIG_H]:
	* src/libs/libgroff/device.cpp [HAVE_CONFIG_H]:
	* src/libs/libgroff/fatal.cpp [HAVE_CONFIG_H]:
	* src/libs/libgroff/string.cpp [HAVE_CONFIG_H]:
	* src/libs/libgroff/strsave.cpp [HAVE_CONFIG_H]:
	* src/preproc/eqn/other.cpp [HAVE_CONFIG_H]:
	* src/preproc/eqn/text.cpp [HAVE_CONFIG_H]:
	* src/preproc/pic/object.cpp [HAVE_CONFIG_H]: Do it.

	* src/libs/libgroff/assert.cpp [HAVE_CONFIG_H]:
	* src/libs/libgroff/errarg.cpp [HAVE_CONFIG_H]:
	* src/libs/libgroff/error.cpp [HAVE_CONFIG_H]: #include
	config.h using angle brackets instead of quotation marks.

	Fixes <https://savannah.gnu.org/bugs/?60035>.

2021-02-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #60025.

	* tmac/tests/doc_Mt-works.sh: Do it.
	* tmac/tmac.am (tmac_TESTS): Run test.

2021-02-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	[ms]: Demote definition of \[yogh] special character escape from
	unconditional availability to be defined only if the output
	device defines it or if .AM macro is called to enable support
	for Berkeley-style accent marks.

	See
	  https://lists.gnu.org/archive/html/groff/2021-01/msg00000.html
	and follow-ups for discussion.

	* tmac/s.tmac (initialization): Move definition of \[yogh] (only
	if the output device doesn't already define it) from here...
	(AM): ...to here.

2021-02-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/ms.ms: Use \[ps] special character instead of local
	character definition using numeric code point escape, which is
	less portable.

	Fixes <https://savannah.gnu.org/bugs/?59528>.

2021-02-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac (initialization): Call par*define-font-macro with
	arguments "CW" and "R" in nroff mode to silence font warnings
	from documents using .CW macro in nroff mode.

2021-02-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/table.cpp (table::init_output): Save value of
	hyphenation maximum consecutive line count register (\n[.hlm])
	more carefully to keep it from being incorrectly parsed as a
	decrement, because negative values of \n[.hlm] are valid (in
	fact, "-1" is the default).  In documents with a large number of
	tables, this can lead to a Persian chessboard problem and an
	integer overflow (at the groff language level, caught and
	handled by src/roff/troff/number.cpp:parse_term()).

	It is wise to wrap a non-literal second argument to the .nr
	request in parentheses if assignment is desired and its value
	can be negative.  See section 5.6.1 ("Setting Registers") of the
	groff Texinfo manual.

	Fixes <https://savannah.gnu.org/bugs/?59993>.

2021-02-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* test-groff.in: Add support for rfc1345.tmac.

2021-02-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	Integrate rfc1345.tmac into build system.

	* contrib/rfc1345/rfc1345.am: Do it.
	* Makefile.am: Include contrib/rfc1345/rfc1345.am.

2021-02-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (open_mac_file): Report problem when
	attempting to open macro files (-m arguments) and the error is
	something other than ENOENT.
	(process_macro_file): Update diagnostic to be more precise; the
	file couldn't be _opened_, not necessarily _found_, and clarify
	that the string being reported back to the user is the argument
	to the -m option, not a file name.
	(macro_source): Update diagnostic to report that the file
	couldn't be _opened_, not necessarily _found_, and include the
	nature of the problem.

2021-01-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	[tbl]: Save and restore hyphenation parameters.

	* src/preproc/tbl/table.cpp (table::init_output): When starting
	a table, save the hyphenation parameters (\n[.hy], \n[.hla],
	\n[.hlm], \n[.hym], \n[.hys]).  Restore them in the table reset
	macro (confusingly called "3init"), which is called before
	performing each diversion and before exiting a table.

	This enables people to, e.g., turn off hyphenation in a table
	text block with .nh, just as they can turn off adjustment with
	.na, without having to manually reset it.  The next text block,
	and the material after the table, will not be affected.

	Fixes <https://savannah.gnu.org/bugs/?59971>.

2021-01-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #59971.

	* src/preproc/tbl/tests/\
	save-and-restore-hyphenation-parameters.sh: Do it.
	* src/preproc/tbl/tbl.am (tbl_TEST): Run test.

2021-01-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tmac.am ($(TMACSTRIPFILES)):
	($(TMACMDOCSTRIPFILES)): Remove unidiomatic for loop which
	frustrated parallel make operation.

	Thanks to an anonymous contributor for the report and patch.

	Fixes <https://savannah.gnu.org/bugs/?59958>.

2021-01-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/ms.ms (Document control settings): Document old FAM string
	and new FR string.

2021-01-29  T. Kurt Bond <tkurtbond@gmail.com>

	[ms]: Add new string FR to hold an expression for computing the
	footnote length relative to the line length, consistently for
	single- and multi-column modes, which \n[FL] has never done.

	* tmac/s.tmac: Initialize FR string to "11/12".
	(@MC): Compute footnote column width using \*[FR] instead of a
	hard-coded "11/12" expression.
	(par@init): Compute default \n[FL] using \*[FR] instead of a
	hard-coded "11/12" expression.

	* doc/groff.texi (Macro Packages/ms/Document control settings):
	* tmac/groff_ms.7.man (Usage/Document control registers):
	Document it.

	Fixes <https://savannah.gnu.org/bugs/?59605>.

2021-01-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/env.cpp: Add ADJUST_MAX enumeration constant to
	record maximum valid numerical adjustment mode.
	(adjust): Verify numeric argument against ADJUST_MAX instead of
	a numeric literal.  Ignore excessively large values instead of
	setting adjustment mode to "right".  Update warning diagnostic.

2021-01-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for .ad and .na requests.

	* src/roff/groff/tests/adjustment_works.sh: Do it.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

2021-01-24  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Manipulating Filling and Adjustment):
	* man/groff.7 (Requests/Request short reference): Clarify
	behavior of ".na" and ".ad l".

	Thanks to Bjarni Ingi Gislason for pointing out the potential
	for confusion and for his careful review of the new text.

	Fixes <https://savannah.gnu.org/bugs/?59795>.

2021-01-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/grog/grog.pl: Report program name in fatal error
	diagnostics.  Also drop sentence-ending punctuation since Perl
	supplies additional information.

	Continues the long process of fixing Savannah #52463.

2021-01-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (RE): Style-warn if macro arguments are out
	of range or redundant.

	Fixes <https://savannah.gnu.org/bugs/?55320>.

2021-01-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	Improve style of troffrc{,-end} files.  Follow established
	idioms and make them consistent with each other.

	* tmac/troffrc:
	* tmac/troffrc-end:
	  - Identify files in header comments as part of GNU troff.

	* tmac/troffrc:
	  - Clarify purpose of .do request.
	  - Identify .X (set by groff -X) as a register, not a string.
	  - Remove temporary registers one per line to make
	    synchronization with foregoing logic easier.  Fixes name
	    space management nits: an undefined string troffrc!Xps was
	    being removed, and troffrc!{ascii,latin1,utf8,cp1047} were
	    not being removed despite being defined.

	* tmac/troffrc-end:
	  - Clarify that file is read after -m file arguments (not "all
	    macro sets", which can be loaded with .mso after this file
	    is processed).
	  - Advise usage of .do for groff extensions.
	  - Line-break one-off conditionals as troffrc does.
	  - Add empty string comment to empty string definitions.

2021-01-14  Dave Kemper <saint.snit@gmail.com>

	Commit 87edb525, from 2003, added character U+2026 (HORIZONTAL
	ELLIPSIS) to most base groff fonts, but there has been no
	kerning information for this character.  To produce consistent
	typography, it should be kerned the same way as the period,
	which is in 818 kern pairs across all the devps fonts.

	Apply the following shell command to the groff description files
	of the PostScript fonts.

	  for file in font/devps/*[A-Z]
	    do sed -Ei\~ 's/(.*)(^| )\. (.*)/&\n\1\2u2026 \3/' $file
	  done

	* font/devps/AB:
	* font/devps/ABI:
	* font/devps/AI:
	* font/devps/AR:
	* font/devps/BMB:
	* font/devps/BMBI:
	* font/devps/BMI:
	* font/devps/BMR:
	* font/devps/HB:
	* font/devps/HBI:
	* font/devps/HI:
	* font/devps/HNB:
	* font/devps/HNBI:
	* font/devps/HNI:
	* font/devps/HNR:
	* font/devps/HR:
	* font/devps/NB:
	* font/devps/NBI:
	* font/devps/NI:
	* font/devps/NR:
	* font/devps/PB:
	* font/devps/PBI:
	* font/devps/PI:
	* font/devps/PR:
	* font/devps/TB:
	* font/devps/TBI:
	* font/devps/TI:
	* font/devps/TR:
	* font/devps/ZCMI: Apply above script.

	Fixes <https://savannah.gnu.org/bugs/?58897>.  However, this
	will need to be done again if afmtodit is used to regenerate the
	above files, or afmtodit will need to be modified to add this
	kerning information itself.

2021-01-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libgroff/relocate.cpp (set_current_prefix) [_WIN32]:
	Move declaration of `pathextstr` to lie within preprocessor
	conditional, since it is dereferenced only there.

2021-01-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add support for strsignal().

	POSIX.1-2008 added strsignal() to the C library and recommended
	its use over sys_siglist[], but groff's pipeline management
	hadn't been updated in that respect since that time.

	* configure.ac: Check for strsignal().
	* src/roff/groff/pipeline.c (xstrsignal): Return strsignal() if
	it is defined.

	Thanks to an anonymous contributor for the report and the patch.

	Fixes <https://savannah.gnu.org/bugs/?59835>.

2021-01-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/libs/libgroff/relocate.cpp (relocatep): Add assertion to
	identify logic error if `curr_prefix` is unexpectedly a null
	pointer.
	(set_current_prefix) [_WIN32]: Allocate memory from heap for
	`curr_prefix` only on Windows; on other systems, this file's
	searchpath() is used to populate `curr_prefix`, and that
	function (except on Windows) performs its own allocation.  Fixes
	memory leak noted by Ingo Schwarze.
	(set_current_prefix) [!_WIN32]: Move logic attempting to set
	`curr_prefix` by calling searchpathext() from here...
	[WIN32]: ...to here.  The PATHEXT environment variable has
	semantics only under Windows, not POSIX systems, so the
	placement of this code seemed erroneous.

	See <https://savannah.gnu.org/bugs/?55475>.

2021-01-06  Colin Watson <cjwatson@debian.org>

	* contrib/glilypond/glilypond.pl:
	* contrib/gperl/gperl.pl:
	* contrib/gpinyin/gpinyin.pl:
	* tmac/hyphenex.pl: Avoid Perl's unsafe "<>" operator.

	The "<>" operator is implemented using the two-argument form of
	"open", which interprets magic such as pipe characters, allowing
	execution of arbitrary commands which is unlikely to be
	expected.  Perl >= 5.22 has a "<<>>" operator which avoids this,
	but also forbids the use of "-" to mean the standard input,
	which is a facility that the affected groff programs document.

	ARGV::readonly would probably also fix this, but I fundamentally
	dislike the approach of escaping data in preparation for a
	language facility to unescape it, especially when the required
	escaping is as non-obvious as it is here.  (For the same reason,
	I prefer to use subprocess invocation facilities that allow
	passing the argument list as a list rather than as a string to
	be interpreted by the shell.)  So I've abandoned this dubious
	convenience and changed the affected programs to iterate over
	command-line arguments manually using the three-argument form of
	open.

	glilypond doesn't need the initial unshift since that's already
	handled in contrib/glilypond/args.pl.

	Fixes <https://savannah.gnu.org/bugs/?55557>.

2021-01-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac: Set footnote line length in multi-column
	environments to 11/12ths of the text line length for consistency
	with earlier change to FL register default.

	Thanks to T. Kurt Bond for bringing this issue to my attention.

2021-01-04  John Gardner <gardnerjohng@gmail.com>

	* tmac/strip.sed: Escape '.' wildcard when matching lines using
	.as and .ds requests.

	Fixes <https://savannah.gnu.org/bugs/?59498>.

2021-01-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	Document use of SOURCE_DATE_EPOCH and TZ.

	The semantics of SOURCE_DATE_EPOCH to groff were not established
	with respect to time zone selection, prompting divergent
	interpretations; Debian and distributions derived from it have
	for several years patched groff to implicitly use UTC as the
	time zone when interpreting the current time (or
	SOURCE_DATE_EPOCH) as a local time.  While a convenient and
	defensible choice for reproducible build efforts[1], it runs
	against the grain of user expectations.  Systems programmers
	like monotonically increasing clocks; the broader user base
	usually prefers a clock that follows an applicable civil
	calendar.  To the latter audience, a difference between
	  $ date "+%Y-%M-%d %H:%m:%S"
	and
	  $ groff <<EOF
	  .af year 0000
	  .af mo 00
	  .af dy 00
	  .af hours 00
	  .af minutes 00
	  .af seconds 00
	  .tm \n[year]-\n[mo]-\n[dy] \n[hours]:\n[minutes]:\n[seconds]
	  EOF
	is difficult to explain.

	* doc/groff.texi (Environment):
	* src/devices/grohtml/grohtml.1.man (Environment):
	* src/devices/gropdf/gropdf.1.man (Environment):
	* src/devices/grops/grops.1.man (Environment):
	* src/roff/groff/groff.1.man (Environment):
	* src/roff/troff/troff.1.man (Environment): Expand description
	of use of SOURCE_DATE_EPOCH to note that it converts the time
	{as seconds since the Epoch} to local time, which motivates
	documentation of the TZ environment variable.  Explain what the
	time stamps are used for: troff merely stores the human-readable
	time components in registers; the grohtml, gropdf, and grops
	output drivers write the information to the output in comments.

	Fixes <https://savannah.gnu.org/bugs/?57218>.

	[1] https://reproducible-builds.org/docs/source-date-epoch/

2020-12-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/doc.am (.texi.dvi): Call texi2dvi with FORCE_SOURCE_DATE=1
	in the environment, avoiding an embedded timestamp in the
	generated groff.dvi file, which frustrated reproducible builds.
	Thanks to Werner Lemberg for the suggestion.

	* src/roff/groff/tests/string_case_xform_unicode_escape.sh: Fix
	test to no longer use Bash process substitution, resulting in
	nondeterministic file descriptor numbers appearing in test logs,
	frustrating reproducible builds.

	* contrib/pdfmark/pdfmark.am (PDFROFF): Call pdfroff without
	`--keep-temporary-files` option.  Temporary directories are
	created with mktemp(1) and files with an embedded process
	identifier, which frustrates reproducible builds.

	See <https://savannah.gnu.org/bugs/?57218>.

2020-12-21  Dorai Sitaram <ds26gte@yahoo.com>

	* tmac/s.tmac (@IP): Handle inter-sentence space correctly in
	paragraph tags by copying \n[.sss] from the enclosing
	environment to that used to format the paragraph tag.

	Fixes <https://savannah.gnu.org/bugs/?59742>.

2020-12-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #59742.

	* tmac/tests/s_IP_respects_inter-sentence_space_in_tags.sh: Do
	it.
	* tmac/tmac.am (tmac_TESTS): Run it.

2020-12-21  Bjarni Ingi Gislason <bjarniig@rhi.hi.is>

	* src/utils/addftinfo/addftinfo.cpp (usage): Use size_t for loop
	index when iterating over a count of size_t items.  Quiets
	signedness mismatch compiler warning.

	Fixes <https://savannah.gnu.org/bugs/?59583>.

2020-12-20  Colin Watson <cjwatson@debian.org>

	* src/devices/gropdf/gropdf.pl:
	* src/utils/afmtodit/afmtodit.pl: Sort Perl hash keys.  Hash
	iteration order may differ between runs, which makes builds
	harder to reproduce.  Sort hash keys in gropdf and afmtodit
	output to avoid this.

2020-12-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/ms.ms (Basic Information): Tweak unit definitions.  groff
	defines a typesetter's point as precisely 1/72 inches.  Also use
	the correct symbol for inch units--strictly, it's the same as
	that for "seconds" as in subdivisions of the degree, and not a
	typographical quote of any sort.

	Fixes <https://savannah.gnu.org/bugs/?59463>.

2020-12-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	[ms]: Set footnote line length to AT&T default.

	groff ms has since its initial implementation used a default
	footnote line length of 5/6ths of the text line length; this may
	correspond to an early AT&T ms default (perhaps documented in
	the 1974 version of the M. E. Lesk paper, "Typing Documents on
	the UNIX System"[1]).  However, as early as V6 Unix (1975), AT&T
	ms actually used a footnote line length of 11/12ths of the text
	line length instead[2].

	Heirloom Doctools and neatroff ms also use this default.

	* tmac/s.tmac: Set default footnote line length to 11/12ths of
	the text line length.

	* doc/groff.texi (ms Document Control Settings):
	* doc/ms.ms (Document control registers): Document new default.

	[1] https://www.troff.org/using-ms.pdf
	[2] https://minnie.tuhs.org/cgi-bin/utree.pl?file=V6/usr/lib/tmac.s

2020-12-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	Lower new unplanted trap error to 'mac' warning.

	* src/roff/troff/div.cpp (top_level_diversion::change_trap):
	Change error upon invalid attempt to move an unplanted trap into
	a warning of type 'mac'.

	* doc/groff.texi (Warnings):
	* src/roff/troff/troff.1.man (Warnings): Recast description of
	'mac' warning type to include the above scenario.

	Thanks to Bjarni Ingi Gislason for reporting the diagnostic
	arising in real life, and to Peter Schaffter for the discussion
	and recommendation.  Some bike sheds get painted quickly!

	Fixes <https://savannah.gnu.org/bugs/?59573>.

2020-12-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_mdoc.7.man: Tweak mandatory macro explanations.

	Update descriptions and template of .Dd, .Dt, .Os usage to
	reflect recent changes and recommended conventions.

	Thanks to Ingo Schwarze, Colin Watson, and Alan D. Salewski for
	the discussion, and Florent Rougon and Robert Bihlmeyer (many
	years ago) for the original report.

	Fixes <https://bugs.debian.org/284002>.

2020-12-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/xtotroff/xtotroff.c (MapFont): Avoid writing past
	the end of a static buffer.  Problem found and patch supplied by
	Bjarni Ingi Gislason.  I tweaked it to comment it differently
	{in case the buffer ever needs to grow, but the prospects of
	future X11 server-side font rendering development seem dim} and
	use snprintf() instead of retaining the existing sprintf().

	Fixes <https://savannah.gnu.org/bugs/?55257>.

2020-12-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/xtotroff/xtotroff.c (CanonicalizeFontName,
	FontNamesAmbiguous, MapFont, main): Format diagnostic messages
	more consistently with GNU Coding Standards.  Prefix with name
	of complaining command.  Put argument literals in quotation
	marks.  Put each messge on one line only.

2020-11-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Parameters): Fix error in example.  Arguments
	to .als were in the wrong order (.als is not ln(1)).  Also mark
	output to the standard error stream using the @error Texinfo
	command instead of @result.

	Thanks to Dorai Sitaram for finding this error.

	Fixes <https://savannah.gnu.org/bugs/?59566>.

2020-11-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Debugging): Update with a more helpful
	introduction, summarizing available procedures.  Mention
	backtracing since it is much more useful now (post-groff
	1.22.4.)

	* man/groff.7 (Debugging): Add new section summarizing
	material added to our Texinfo manual.

	* man/groff_diff.7.man (Debugging): Add new section describing
	groff extensions.

2020-11-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Traps): Update.  Organize subsubsections "Page
	Location Traps" and "Diversion Traps" under new subsection node
	"Vertical Position Traps" to make the conceptual organization
	more clear.  Define and discuss trap visibility earlier.
	Document unit used for page location trap reporting.  Introduce
	analogy of .wh and .ch requests as queue operations.  Document
	what happens when the same macro is planted as a trap in
	multiple locations; supply example.  Supply example of .itc
	usage.  Consistently refer to an end-of-input macro as such, to
	contrast it with the "end macro" that can be used with, e.g.,
	.de and .ig requests.

	* man/groff.7 (Traps): Add new section summarizing
	language feature.

2020-11-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/mdoc/doc-common-u (doc-footer): When performing
	double-sided page layout, set page number on outside of _recto_
	{odd-numbered} pages, as is conventional and for consistency
	with man(7).

	* tmac/groff_mdoc.7.man
	(Formatting with groff, troff, and nroff): Document what
	double-sided layout (not "printing") means.  Also correct
	description of continuous rendering, which implied that page
	headers and footers were not written at all, and used an
	incorrect groff driver name.

	Fixes <https://bugs.debian.org/919890>.  Also see:
	https://lists.gnu.org/archive/html/groff/2019-01/msg00021.html
	Thanks to Ralph Corderoy for the discussion.

	* tmac/tests/doc-D_places_page_numbers_correctly.sh: Test it.
	* tmac/tmac.am (tmac_TESTS): Run test.

2020-11-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/mdoc/doc-common-u (doc-header): Fix infinite loop when
	attempting to trim header string (like "FTP(1)") to fit when the
	available title length won't allow even extreme shortening.
	Measure the string before and after calling .substring on it,
	and break out of loop if it didn't get shorter.

	Problem dates back to at least
	ed63b0ae76a611b581601a1afc192f6a7367be6f (7 July 2002), possibly
	as far as the Great Mdoc Rewrite of
	058f72af832fc68488d33cd09ec819e5c560fa09 (23 March 2001).

	One can never check loop invariants too many times...

	Fixes <https://bugs.debian.org/411227>.

	* tmac/tests/\
	doc-do_not_loop_infinitely_when_shortening_headers.sh: Test it.
	* tmac/tmac.am (tmac_TESTS): Run test.

2020-11-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/nroff/nroff.sh: Recognize -b and -z options and pass
	them through to troff.  Document them in usage message.
	* src/roff/nroff/nroff.1.man: Document new -b and -z options.

2020-11-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/div.cpp (top_level_diversion::change_trap):
	Emit error diagnostic if an attempt is made to move an unplanted
	macro.  This could have been a warning, as it's pretty harmless
	{though possibly a big surprise to anyone who was wondering why
	their .ch was a no-op}, but there's no good warning category for
	this kind of problem and I am not about to start a bike shed
	discussion about it right now.

2020-11-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi: Update.  Add introductory material.  Rewrite
	the first section of the "gtroff Reference" chapter of our
	Texinfo manual.  It is written as an introduction for readers
	who want to go straight to "raw" troff without knowing much or
	anything about existing macro packages.  Thanks to Dave Kemper
	and John Gardner for feedback and support.

	Clarify whitespace usage.  "Whitespace" is defined in this
	manual as "spaces, tabs, and newlines".  Say only "spaces and
	tabs" when newlines should not be included.

	Rename "Font Files" to "Device and Font Files".

	Rename "Manipulating Filling and Adjusting" to "Manipulating
	Filling and Adjustment".

	Update discussion of "copy mode".  Rename from "copy-in mode",
	which I don't think eludicdated anything; is there a "copy-out
	mode"?  Rename nodes accordingly.  Attempt to explain more
	clearly.  Recast to shift emphasis to what _isn't_ merely copied
	in copy mode, since that is what seems to cause confusion among
	the inexperienced.

	Update "Conditionals and Loops".  Add introductory paragraph.
	Add nodes/subsections "if-then" and "Conditional Blocks".  Add
	subsection "Conditional Blocks" to explain the behavior of the
	\{ and \} escapes much more precisely.  I don't think this
	syntactical area is well understood.  Supply examples.

2020-11-14  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add style checks to man(7) macro package.

	Not otherwise documented yet; experimental--subject to change.

	* tmac/an-old.tmac: Initialize CHECKSTYLE to false if not
	already set.
	(an-style-warn): New macro emits diagnostic of type "style" when
	called if CHECKSTYLE register is true.
	(TH): Call an-style-warn if fewer than two or more than five
	arguments are seen.
	(RI, IR, IB, BI, RB, BR): Call an-style-warn if fewer than two
	arguments are seen.

2020-11-11  Bertrand Garrigues <bertrand.garrigues@laposte.net>

	Update copyright.

	* update-copyright.sh: use gnulib's 'update-copyright' script.
	Pass this script in directories 'arch', 'contrib', 'font',
	'man', 'tmac', 'src' and on a list of extra files.

	* FOR-RELEASE: mention this point.

2020-11-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/preconv/preconv.1.man (Description): Fix error:
	groff's -K option, not -k, specifies a character econding (and
	overrides GROFF_ENCODING in the environment).  Also explain how
	valid encoding strings are determined.

	Thanks to Bjarni Ingi Gislason for the report.

	Fixes <https://savannah.gnu.org/bugs/?59445>.

2020-11-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/table.cpp (table::compute_separation_factor):
	Add word "table" to diagnostic message to better cue the user
	that it is produced by roff that has been injected into the
	document by the tbl(1) preprocessor.  The other 3 such possible
	diagnostic messages already do this.  Thanks to Bjarni Ingi
	Gislason for bringing the inconsistency to light.

	Fixes <https://savannah.gnu.org/bugs/?59443>.

2020-11-07  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_mdoc.7.man (Predefined strings): Ensure the table
	fits even on ASCII and Latin-1 terminals.  Thanks to Bjarni Ingi
	Gislason for the report.

	Fixes <https://savannah.gnu.org/bugs/?59424>.

2020-11-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_mdoc.7.man (General text domain/Enclosure and
	Quoting Macros): Fix errors in macro descriptions.
	<.Dq>: Encloses its arguments in directional double quotes where
	available.
	<.Eq>: Misrendered example; say "XstringY" rather than
	"XXstring".
	<.Sq>: Encloses its arguments in directional single quotes where
	available.

2020-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add support for CS and CT registers to mdoc(7), just like in
	man(7), from a year ago.

	* tmac/doc.tmac-u (doc-print-recursive): Call .stringup on each
	argument if register doc-do-capitalize is true.
	* tmac/mdoc/doc-common-u (Dt): Call .stringup on
	doc-document-title if \n[CT] is true.
	(Sh): Set doc-do-capitalize to value of \n[CS].  Set
	doc-do-capitalize false before returning.
	* tmac/mdoc/doc-ditroff-u:
	* tmac/mdoc/doc-nroff-u: Set CS and CT registers to 0 (false)
	if the user has not defined them.

	* tmac/groff_mdoc.7.man \
	(Formatting with groff, troff, and nroff): Document it.

	* tmac/tests/doc-CS_works.sh:
	* tmac/tests/doc-CT_works.sh: Test it.
	* tmac/tmac.am (tmac_TESTS): Run tests.

2020-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (do_overstrike, do_bracket,
	do_name_test, do_expr_test, do_zero_width, do_width, do_special,
	do_if_request, read_color_draw_node): Improve diagnostic
	messages for missing closing delimiters by describing the
	incomplete structure and the problem token.
	(read_rgb, read_cmy, read_cmyk, read_gray): Improve English
	syntax of diagnostic message when color definition missing.

2020-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/mdoc/doc-ditroff-u:
	* tmac/mdoc/doc-nroff-u: Recognize but ignore the groff_man(7)
	string HF and registers FT, IN, P, SN, and X (by initializing
	them empty or zero if they are not set).  This reserves them so
	that they don't become used for divergent purposes.  man(1)
	programs set these and other parameters already handled (like
	LL) to configure page rendering, and it would break the
	macro-package agnosticism afforded by andoc.tmac to expose
	different externally-programmable registers and strings.
	* tmac/groff_mdoc.7.man \
	(Formatting with groff, troff, and nroff): Document this.

2020-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac: Implement \*< and \*> strings for subscripting.
	groff ms has had (extension) strings for superscripting since
	1991 at the latest, and the asymmetry possessed me of a madness.
	Both pairs of string names follow similar usage in Eric Allman's
	"me" macros.

	* NEWS:
	* doc/groff.texi:
	* tmac/groff_ms.7.man: Document it.

2020-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_mdoc.7.man (Predefined strings): Fix error: the
	\*[Lq] and \*[Rq] strings degrade to neutral double quotes (")
	on "nroff" devices (-Tascii and -Tlatin1), not "``" and "''".

2020-10-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac (ds*end!0): Improve diagnostic to mention .RD.

2020-10-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/s.tmac: Revise diagnostic messages.  Define a common
	prefix string, starting all such messages with "s.tmac".
	Continues the long process of fixing Savannah #52463.  Drop the
	word "macro" from diagnostic since this will now be clear from
	the prefix.  Prefix every macro name in a diagnostic message
	with a leading dot; this was already done in four cases.
	(cov*first-page-init): Use temporary string to avoid multi-line
	diagnostic message.
	(@NH): Use temporary string to prevent overlength line.

	A further benefit of this revision is that index information
	produced by ms's .IX macro (which writes to the standard error
	stream) will now be trivially easy to extract even for documents
	that cause diagnostic messages.  One can simply filter them with
		grep -v '^s\.tmac:'
	or similar.

2020-10-28  Ingo Schwarze <schwarze@openbsd.org>

	* man/roff.7.man, tmac/groff_man.7.man.in: Correct man(7)
	history.

2020-10-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	Fix ms .R macro to work as documented, by handling its arguments
	instead of ignoring them.

	* tmac/s.tmac (R): Delete.  During set up, call
	par*define-font-macro for R font just as we do for B, I, and BI.

	Problem dates back at least to groff 1.02, June 1991.

	* tmac/tests/s_R-handles-its-arguments.sh: Test it.
	* tmac/tmac.am (tmac_TESTS): Run test.

	Fixes <https://savannah.gnu.org/bugs/?59348>.

2020-10-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	Fix ms .TC macro by allowing it to actually use lowercase Roman
	numerals for the page(s) of the table of contents when a custom
	title (header or footer) is defined containing '%'.

	As a side effect, eliminate the PN register, which was
	apparently misused because it was undocumented and had a name
	just long enough to be both suggestive and ambiguous.  Another
	pin goes into my Ken Thompson voodoo doll.

	* tmac/s.tmac (IX): Write out page number of index term(s) using
	the % register (instead of PN), obtaining whatever format is
	assigned to the register at that time.
	(CH): Similarly, output %, not PN, in the center of the default
	heading.
	(PT): Save the format of %, set it to decimal, copy its value to
	a new register, pg*page-number-in-decimal, then restore %'s
	previous format.  Compare the new register, not %, to 1 to
	determine whether special page one behavior should be used.
	(pg*end-page): Assign pg*next-format to %, not PN.
	(XA): Define toc*num with the interpolation of %, not PN.

	* tmac/tests/s_TC-works-with-percent-in-custom-titles.sh: Test.
	* tmac/tmac.am (tmac_TESTS): Run test.

	Problem appears to be very old; as I read it, Larry Kollar was
	complaining of it in his ms.ms document over 20 years ago.

	Fixes <https://savannah.gnu.org/bugs/?59345>.

2020-10-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff.7 (Requests/Request short reference) <.af>: Fix
	error; "l" is not a valid register format.  Explain what the
	request does, and the default format, as tersely as possible.

2020-10-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	Use only malloc() and free() to manage memory of paths opened by
	the parser instead of mixing in C++ new/delete management under
	some runtime-dependent circumstances.

	* src/libs/libgroff/relocate.cpp (relocatep): Use malloc(), not
	new.
	* src/roff/troff/input.cpp (process_macro_file,
	process_startup_file, macro_source): Use free(), not (a_)delete.

	Thanks to an anonymous contributor for the report and patch.

	Fixes <https://savannah.gnu.org/bugs/index.php?56694>.

2020-10-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff_char.7.man (Description/Special character escape
	forms): Clarify discussion of Unicode Normalization Form D and
	its applicability to code points acceptable in Unicode numeric
	special character escapes.  Also document that these escapes
	must use uppercase hexadecimal digits.

	Attempts once again to slay the unkillable beast that is
	<https://savannah.gnu.org/bugs/index.php?57618>.

2020-10-21  Ingo Schwarze <schwarze@openbsd.org>

	On systems without NAME_MAX, use FILENAME_MAX as a last resort.

	Problem reported by Eli Zaretskii <eliz at gnu dot org>
	on MS Windows in https://savannah.gnu.org/bugs/?55449

2020-10-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/env.cpp (environment::possibly_break_line):
	Emit break warning and return if the output width is not
	positive.  The code assumes that it will be and loops infinitely
	if it isn't.  I _think_ this is because we're not able to get
	width data for (some?) CJK glyphs.  Based on a patch by Osamu
	Sayama.

	* src/roff/groff/tests/\
	do_not_loop_infinitely_when_breaking_cjk.sh: Test it.
	* src/roff/groff/groff.am: Run test.

	Fixes <https://savannah.gnu.org/bugs/index.php?44018>.

2020-10-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/table.cpp (table::init_output): Save the value
	of \n[.tabs] when starting a table.  In the reset macro, restore
	the saved value.

	Based on a patch by Bjarni Ingi Gislason (tweaked to use a more
	normative preprocessor symbol; "REG"s aren't "NAME"s).

	* src/preproc/tbl/tests/save-and-restore-tab-stops.sh: Test it.
	* src/preproc/tbl/tbl.am (tbl_TESTS): Run test.

	Fixes <https://savannah.gnu.org/bugs/index.php?42978>.

2020-10-20  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/nroff/nroff.sh: Recognize -E option and pass it
	through to troff.  Document it in usage message.
	* src/roff/nroff/nroff.1.man: Document new -E option.

	Based on a patch by Ingo Schwarze.

	Fixes <https://savannah.gnu.org/bugs/index.php?44289>.

2020-10-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Stop remapping ` and '.  Our own pages now
	appear to be clear of wrong-quote problems, so let's make them
	visible if they recur.  Those who don't want to fix bad man
	pages (distributors, site admins) can restore the mappings in
	their man.local files.

2020-10-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-warn): New; emits warning diagnostic.
	(R): Use new macro instead of .tm* requests directly.

2020-10-18  Ingo Schwarze <schwarze@openbsd.org>

	#include "config.h" before <stdio.h>

	Required with e.g. gcc 4.2.1 because gnulib/lib/stdio.in.h
	uses the "restrict" keyword since this gnulib commit:
	commit 182afcba2635cbff91240656c7fb3742dd23ab6f
	Author: Bruno Haible <bruno@clisp.org>
	Date: Sat Feb 22 20:57:30 2020 +0100

	Otherwise, the build may die from the declaration of
	various printf-like functions with messages like:
	./lib/stdio.h:851: error: expected ',' or '...' before 'fp'

	* src/libs/libgroff/assert.cpp, src/libs/libgroff/errarg.cpp,
	src/libs/libgroff/error.cpp, src/preproc/eqn/eqn.ypp:
	#include "config.h".

2020-10-18  Bertrand Garrigues <bertrand.garrigues@laposte.net>

	hpftodit: incorrect 'delete' after new[]

	* src/utils/hpftodit/hpftodit.cpp (output_font_name): use
	'delete[]'.

	Fixes bug #55331. Found by David Binderman, fixed by Ingo
	Schwarze.

2020-10-18  Bertrand Garrigues <bertrand.garrigues@laposte.net>

	preconv: don't use libuchardet if input is stdin

	* src/preproc/preconv/preconv.cpp (do_file): don't call
	detect_file_encoding if input file is "-"

	This fixes the failure on MS-Windows described #55334, however
	this does not fix the encoding detection with uchardet if the
	input is stdin (the user would have to pass with -D the correct
	encoding as explained in the man page).

2020-10-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/eqn/lex.cpp (troff_defs): Set the "..." token on
	the baseline, not vertically centered, aligning the actual
	behavior with what our eqn(1) man page has claimed since 2007.

	Fixes <https://savannah.gnu.org/bugs/index.php?59285>.

2020-10-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/devices/grops/grops.1.man
	(Usage/TrueType and other font formats): Remove dead URL to
	ttftot42 utility.  Update URL to fontforge tool.  Replace much
	of the discussion with an example, motivated by Jordan Torbiak's
	"groff-install-font" script on GitHub Gist, of how to add the
	Roboto Slab Regular font to a user-local groff font directory
	{and how to test it, too}.

	Fixes <https://savannah.gnu.org/bugs/index.php?58077>.

2020-10-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/afmtodit/afmtodit.pl: Report program name in warning
	diagnostics.

	Continues the long process of fixing Savannah #52463.

2020-10-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Add fallback for lq, rq strings.

	Define \*(lq and \*(rq as '"' if the output device has no \(lq
	or \(rq special character escapes defined (respectively).

	This is a bit belt-and-suspenders for groff since our output
	devices all guarantee availability of these glyphs, but if this
	macro file gets used with other implementations (and if they
	support the 'c' conditional), it should prevent the strings from
	producing empty output.

	Man page writers should simply use \(lq and \(rq unless they are
	aiming for pathological levels of portability (e.g., composing
	man pages today for consumption on simulators of 1980s Unix
	systems).

2020-10-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Define an-end with .de1.  We need to define
	an-end with de1 (execute macro with compatibility mode off)
	because, as of 37b4180a27a6eeaea429e40ec278abefcda7f3a7 (11
	October), it can now be called from a trap executing in a
	context whence compatibility mode might be on (namely,
	"reload-doc" in andoc.tmac).

2020-10-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/eqn/main.cpp (main):
	* src/preproc/pic/main.cpp (main):
	* src/preproc/tbl/main.cpp (main):
	* tmac/eqnrc:
	* tmac/troffrc: Perform checks of register and macro definitions
	with a .do request, since we might be in compatibilty mode.

2020-10-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (do_if_request): Emit warning of type
	"syntax" if a groff conditional expression extension is used
	when compatibility mode is active.

	* doc/groff.texi:
	* src/roff/troff/troff.1.man: Update description of "syntax"
	warning type to be more general.  It was inaccurate, referring
	only to "dubious syntax in numeric expressions", which was not
	the case.  Instead it was being used only for poor construction
	of character classes with the .class request.

2020-10-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/main.cpp (main): Stop ignoring -T.  GNU tbl
	was undocumentedly ignoring the -T option; apparently IRIX tbl
	implemented it, something on the system called tbl with that
	option (I'm guessing IRIX's man(1)) and its users spammed James
	Clark with bug reports.  We can probably weather the volume of
	such spurious reports from IRIX users today.

2020-10-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/main.cpp (main): Clarify fatal diagnostic in
	generated document by informing the user that the program is
	aborting, and also that it is groff extensions that are required
	rather than GNU troff per se (since Heirloom Doctools troff
	claims groff extension support via the .g register).

2020-10-14  Ingo Schwarze <schwarze@openbsd.org>

	In groff(1), fix the combination of -v with -k, -j, and -J.

	* src/roff/groff/groff.cpp: Pass the -v option through to the
	preconv, chem, and ideal preprocessors, just like for all
	other preprocessors, to print the version of the preprocessor
	and to avoid garbage output.

	Bugfix patch from Eli Zaretskii <eliz at gnu dot org>
	submitted in: https://savannah.gnu.org/bugs/?55297

2020-10-12  Bertrand Garrigues <bertrand.garrigues@laposte.net>

	Update 'gnulib' submodule.

	* gnulib: now points on d60a35e94c4f5b8f09f15828242418f5073d46e7
	from 'gnulib' repository.

	* configure.ac: minimum autoconf version is now 2.64 due to
	gnulib upgrade.

	* FOR-RELEASE: add the gnulib update to the checklist.

2020-10-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/andoc.tmac: When switching between macro packages in
	andoc and using continuous-rendering mode, flush any partially
	collected output line and write page footer.
	(reload-doc): Call an-end if it is defined.
	(reload-man): Call doc-end-macro if it is defined.
	* tmac/tests/andoc-flush-between-packages.sh: Add regression
	test.
	* tmac/tmac.am (tmac_TESTS): Run test.

	Fixes <https://savannah.gnu.org/bugs/index.php?59106>.

2020-10-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/andoc.tmac: Remove traps set by mdoc package by the names
	they actually use.
	* tmac/tests/andoc-clear-doc-traps.sh: Add regression test.
	* tmac/tmac.am (tmac_TESTS): Run test.

	Fixes <https://savannah.gnu.org/bugs/index.php?59246>.

2020-10-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/table.cpp (table::init_output): In the release
	macro written to the output, reword the warning diagnostic that
	is emitted when a table row is to be output without enough room
	before the next page location trap.  The former wording said
	that a "text block" would not fit on the page, and that is
	normally how table rows grow longer than one line in the first
	place.  However, there are other ways the conditional can be
	satisfied, as witnessed in Savannah #57665, so simply say what
	is known; that the table _row_ overruns the space to the next
	trap (we say the row won't "fit on [the] page").

	Fixes <https://savannah.gnu.org/bugs/index.php?59233>.

2020-10-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/table.cpp
	(double_line_entry::double_line_entry):
	(double_line_entry::simple_print): Remove garbage lines
	inadvertently added in previous commit.

2020-10-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	Stop tbl from injecting spurious blank lines (and emitting
	spurious warnings about tables and "text blocks" not "fitting on
	a page") into long tables in man pages when continuous rendering
	is used (the default for nroff output devices).

	* src/preproc/tbl/table.cpp (USE_KEEPS_REG): New preprocessor
	symbol stores name of groff register for dynamic determination
	of keep usage.
	(table::init_output): If the NOKEEP flag is not set, then in
	generated groff output, see if the usekeeps register is defined;
	if not, define it and enable usage of keeps.
	(table::do_row): In groff output, check usekeeps register before
	calling keep and release macros.
	(table::do_bottom): In groff output, check usekeeps register
	before calling release macro.

	* tmac/an-old.tmac (TS): Set the usekeeps register used by tbl
	to the logical complement of the man(7) cR (continuous
	rendering) register.

	* tmac/tests/an-old_TS_do_not_keep_tables_when_cR_set.sh: Add
	regression test.
	* tmac/tmac.am (tmac_TESTS): Run test.

	Problem appears to date back to the introduction of continuous
	rendering in groff 1.17 (3 May 2001).

	Fixes <https://savannah.gnu.org/bugs/index.php?57665>.

2020-10-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/table.cpp (simple_entry::position_vertically):
	(block_entry::position_vertically):
	(table::determine_row_type):
	(printfs): Update assertions to indicate what actually went
	wrong.

2020-10-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/node.cpp: Make diagnostics slightly more
	informative in unusual error cases.
	(real_output_file::~real_output_file): If ferror() reports error
	status on a stream say that, instead of "error writing".  If it
	does not, but fflush() fails on the stream, describe the flush
	as failing, and use strerror() since fflush() sets errno.  If
	pclose() fails, say that we were unable to close a pipe instead
	of repeating the name of the C library function to the user, who
	might not be a C programmer.  Report strerror() in this case and
	for a failing fclose().
	(real_output_file::flush): Repeat updated fflush() logic from
	previous function.

2020-09-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (register setup): Make interaction of \n[C]
	register and HTML output more clear in code and comments.

2020-09-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/tests/regression_savannah_59202.sh: Reduce test
	case.

2020-09-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	Fix SEGV arising from recursing destructor.

	* src/roff/troff/node.h (output_file): Add class member
	`is_dying` to track whether destructor has already been entered;
	initialize false.
	* src/roff/troff/node.cpp (real_output_file::~real_output_file):
	Set `is_dying` true when destructor entered.
	* src/roff/troff/div.cpp (cleanup_and_exit): Only delete
	`the_output` object if it is not already being destroyed.
	* src/roff/groff/tests/regression_savannah_59202.sh: Add test.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

	Thanks to "hackerb9" for reporting the problem.  Problem appears
	to date back to groff 1.02 (June 1991) or earlier.

	Fixes <https://savannah.gnu.org/bugs/index.php?59202>.

2020-09-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (register setup): Condition decisions on
	\n[an-html] rather than a string comparison using \*[.T]; that
	is what the register is there for.

2020-09-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/andoc.tmac: Save alias to .ne request.
	(reload-doc): Restore .ne request, which an-old.tmac meddles
	with if continuous rendering is used.  (The meddling ultimately
	dates back to 777e2d262862621966c18d685a000cc88f432bc6, 26
	January 2002, which simply redefined .ne as empty on nroff
	devices.)

2020-09-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	Ship tests in distribution archive.

	* src/preproc/preconv/preconv.am (EXTRA_DIST):
	* src/roff/groff/groff.am (EXTRA_DIST):
	* src/roff/nroff/nroff.am (EXTRA_DIST):
	* tmac/tmac.am (EXTRA_DIST): Add
	$({preconv,groff,nroff,tmac}_TESTS), as appropriate.

	* src/roff/nroff/nroff.am (MOSTLYCLEANFILES): Stop adding
	$(nroff_TESTS), which can only have unhappy consequences.

2020-09-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tmac.am: Fix problem with recently-relocated mdoc macro
	files not ending up in the right place in the distribution
	archive.
	(dist_tmac_DATA): Also include $(TMACUNSTRIPFILES).
	(TMACMDOCSTRIPFILES): Rename from $(TMACMDOCFILES).  Update
	interpolation sites.
	(TMACMDOCUNSTRIPFILES): New variable holds mdoc macro files in
	the tmac/mdoc subdirectory.  Stop adding these files to
	$(TMACUNSTRIPFILES) since those reside in tmac/ directly.
	(EXTRA_DIST): Add $(TMACMDOCUNSTRIPFILES).
	($(TMACSTRIPFILES)): Depend only on $(TMACUNSTRIPFILES).
	($(TMACMDOCSTRIPFILES)): Add new rule, much like the foregoing.
	Depend on $(TMACMDOCUNSTRIPFILES) and copy files into tmac/mdoc.
	(dist-hook, tmac-dist-hook): Remove; they no longer do any
	distinct work.

	Fixes <https://savannah.gnu.org/bugs/index.php?59186>.

2020-09-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/tbl/table.cpp (table::init_output):
	(table::compute_expand_width):
	(table::compute_separation_factor): Use consistent format for
	diagnostic messages.
	  - Do not spread a single diagnostic across multiple lines.
	  - Report messages in GNU Coding Standards format, that is:
	  - Report the name of the file the problem is in...
	  - ...then the line number...
	  - ...then the diagnostic severity level...
	  - ...then the specific problem.

2020-09-26  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Emit warnings when configuration registers
	are ignored when producing output for HTML.

2020-09-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (an-p-footer): Define \*[an-page-string] if
	(1) the output device is not HTML; (2) \n[X] has been defined;
	and (3) the page number has not yet reached the threshold
	defined by \n[X].

	Fixes <https://savannah.gnu.org/bugs/index.php?59179>.

2020-09-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for Savannah #59179.

	* tmac/tests/an-old_X_register_works.sh: Do it.
	* tmac/tmac.am (tmac_TESTS): Run it.

2020-09-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac:
	* tmac/doc.tmac-u: Emit more informative fatal diagnostic when
	installed version of groff is too old.  Report the version found
	and explicitly note consequent abort.

2020-09-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tmac.am: Regenerate stripped macro files upon changes to
	any of their unstripped counterparts.  This is crude, but they
	weren't being regenerated at _all_.  Defeating a central
	function of Make is bad.
	(TMACUNSTRIPFILES): New variable.  Populate with -u files.
	($(TMACMDOCFILES) $(TMACSTRIPFILES)): Depend on
	$(TMACUNSTRIPFILES).

2020-09-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac:
	* tmac/andoc.tmac:
	* tmac/doc.tmac-u:
	* tmac/s.tmac: Clarify fatal diagnostics by informing the user
	that the program is aborting.

2020-09-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/www.tmac.in: Revise diagnostic message handling.
	(www:fatal): New macro handles fatal macro usage errors.
	(www:lenstr): Use www:fatal instead of writing to standard error
	with .tm (rather than www:error) and then calling .ab without
	arguments, which produces an unsightly "User Abort." message.
	(www:lenstr):
	(www:splitstr):
	(www:url_breaks):
	(www:url_breaks_split):
	(LINKSTYLE): Report expected number of arguments in diagnostics.
	(www:fontstyle): Shorten diagnostic message.

2020-09-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/div.cpp
	(top_level_diversion::clear_diversion_trap): Fix copy and paste
	error in diagnostic, which wrongly reported that a top-level
	diversion trap couldn't be "set" when "clear"ed was meant.

2020-09-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/grn/main.cpp (conv):
	* src/roff/troff/node.cpp (suppress_node::tprint): Remove
	embedded newlines from diagnostic messages.

2020-09-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Move setup of remaining rendering option
	parameters (registers C, D, P, and X) to end of file.

2020-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (TH): Move setup of rendering parameter
	registers IN and SN from here to the end of the file.  Update
	comments.

2020-09-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Drop \*[an-empty] string.  This string is
	documented as preventing "looping" (presumably infinitely) if
	someone calls a two-font macro with an inconvenient parameter;
	the example shown is ".RB ( \\ )".  This string was being
	interpolated at the end of each argument pair to the two-font
	macros BI, BR, IB, RB, and RI.  This code dates back to groff
	1.10 (dc5351364982f78f8c630f1e856d692d4a82666f, 26 November
	1995).  Did you notice a two-font macro missing?  It was: IR.
	Using it with the proscribed input fails to cause a problem; the
	others similarly operate just fine when the empty string
	interpolation is removed.  Presumably at some point in the past
	25 years, this workaround became unnecessary.
	(BI, BR, IB, RB, RI): Remove interpolations of string.

	* tmac/tests/an-old_avoid_two-font_denial_of_service.sh: Add
	regression test.
	* tmac/tmac.am (tmac_TESTS): Run test.

2020-09-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Refactor continuous rendering handling.
	Relocate set up of \n[cR] to end of file in parallel with other
	rendering parameter management.  Instead of conditionally
	defining the an-ne and an-bp macros, define them
	unconditionally.  Also define them with .de instead of .de1,
	since they will only be called within the contexts of macros
	that are already running with compatibility mode off (i.e., the
	public interface macros).  Relocate definition of an-end to sit
	alongside these other continuous-rendering-mode specific macros
	and avoid a forward reference and scattering of \n[cR]
	conditionals around the file.
	(an-set-up-continuous-rendering): Define new macro to move
	macros (and end macro) into place.  Call it at the end of the
	file only if cR eventually winds up being true.

2020-09-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Improve name space management.  Rename .ne
	request to .an-ne instead of outright clobbering it (and in
	parallel with .an-bp).  Rename 'an-ne' register to 'an-need' to
	prevent confusion with renamed request.

2020-09-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Drop .ll hack for LL register.

	Drop the .ll hack for setting the line length on nroff devices.

	Once upon a time, the only way to get man pages to render on a
	terminal at any width other than the nroff default of 65n was to
	put an .ll request into the page--a mortal sin against
	portability--or your man(1) program could sneak such a request
	into nroff's input stream.

	Also, long ago, John Eaton of UT Austin wrote a man(1) program.
	In a few years this implementation branched into two lines of
	development, which I'll call man-db man (Wilford/Polacco/Watson)
	and another, Brouwer/Lucifredi man, which as of this writing saw
	its last release in 2011 (1.6g).

	The man and mdoc macros of groff 1.18 (July 2002) introduced an
	LL register to configure the line length, respecting an -rLL
	command-line option.  The source change was made by Werner
	Lemberg on 3 May 2002 and Colin Watson updated man-db man(1)--on
	the same day!--to synthesize the option when calling groff.

	Later, in August 2005, Keith Marshall, a user of vesion 1.5m of
	Brouwer/Lucifredi man(1), which did not set the LL register,
	suggested that the existing .ll setting (technically, the value
	of \n[.l], the only way the result of an .ll request could be
	introspected) be honored in the absence of the LL register.
	However, it is impossible to distinguish a user-supplied ".ll
	65n" request from nroff's default setting, which was in that
	case overridden to the modern default of 78n.

	And so, in what is now git commit
	f9d5df4aebd3d834b4084ffefa52a115e00dce38 (1 September 2005), it
	was done.

	This led to (1) surprising behavior for users accustomed to old
	methods and desirous of the nroff default and (2) lengthy
	apologia in groff source code comments and the groff_man(7) man
	page.

	Ironically, Brouwer/Lucifredi man(1) had already added support
	for the LL register by the time of its 1.6 release two months
	earlier (20 June 2005).  (Curiously, it did so similarly to the
	old .ll hack, by injecting an '.nr LL' request into groff's
	input stream, rather than using the -r command-line option.)
	Moreover, Marshall and the groff list were already aware of
	this, but the change was accepted anyway because version 1.6
	was "too new"!  (Distributions are slow sometimes, true...)

	Let us survey the field in 2020.  man-db man(1) has supported
	the LL register for eighteen years, and Brouwer/Lucifredi man(1)
	for fifteen.  Heirloom Doctools's man macros set the line length
	to 78n on nroff devices unconditionally.  mandoc(1) similarly
	also always formats for 78 columns on terminals.  groff's
	mdoc(7) macros grew support for LL in parallel with man(7) in
	2002 and never added the \n[.l] introspection at all.

	There no longer seems to be any reason to preserve this hack.

	* tmac/an-old.tmac (initialization): Drop complex setup of LL
	register.
	(TH): Relocate line length reset; no behavior change.
	(after .mso man.local): Initialize LL only if the user hasn't;
	use device default in troff mode, and 78n in nroff mode.

	* tmac/groff_man.7.man.in (Options) <-rLL>: Stop documenting .ll
	hack.

	Fixes <https://savannah.gnu.org/bugs/index.php?58992>.

2020-09-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tests/an-old_LL_init_sanely.sh: Add test.
	* tmac/tmac.am (tmac_TESTS): Run test.

2020-09-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/devices/xditview/GXditview.ad: Widen the default geometry
	to accommodate the width of a page rendered using the X100-12
	device.  Display devices these days have much greater horizontal
	resolution than they did 20 years ago.

2020-09-14  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Use correct point size default for
	X{75,100}-12 devices.

2020-09-14  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/mdoc/doc-common-u (doc-prepare-section-heading): New
	macro defines new string doc-sec-head to enable recognition of
	mixed-case section headings in mdoc man pages.  For example,
	"Name" and "Description" are now recognized in addition to
	"NAME" and "DESCRIPTION".
	(doc-section-{name,synopsis,library,description,see-also,files,
	authors}): Redefine strings to be mixed-case.  Add trailing
	comment guard per recommended best practice.
	(Sh): Call doc-prepare-section-heading instead of
	doc-first-parameter, and compare predefined section strings to
	doc-sec-head instead of doc-str-dfp.

	* tmac/tests/doc-accept-mixed-case-section-headings.sh: Test it.
	* tmac/tmac.am (tmac_TESTS): Run test.
	* tmac/tests/doc-smoke-test.sh: Use mixed-case section headings.

2020-09-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* makevarescape.sed: Use \[] form of special character escapes.

	The only man page we have that doesn't turn compatibility mode
	off (neqn) also doesn't use any sed-substituted patterns where
	characters replaced by a \[] special character escape form are
	likely to be used.

	The @g@ command prefix is the main avenue for intrusion of such
	escapes, but I think it improbable that many people are going to
	include apostrophes, double quotes, carets, grave accents, or
	tildes in the command prefix; some or all of these will pick
	fights with the shell and require quoting that we don't
	represent in man page text anyway.

	On top of that, neqn is largely a stub page.

	If it's a problem, a better fix than reverting this is to make
	neqn switch out of compatibility mode like our other man pages.

2020-09-03  G. Branden Robinson <g.branden.robinson@gmail.com>

	* FOR-RELEASE: Start a release checklist.  We probably should
	have done the afmtodit.tables update for 1.22.4.  There were no
	AGL changes and the Unicode decomposition changes were limited
	to additions for Balinese (11), CJK compatibility ideographs
	(5), and 13 code points outside the Basic Multilingual Plane.

2020-09-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/afmtodit/afmtodit.tables: Regenerate using Unicode
	13.0.0.

2020-09-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	make-afmtodit-tables: Automate AGL reassignments.

	Automate the procedure done manually in
	b72b91e85e80a69304e6569db535bcca0e0ecab5 (9 April 2007), so that
	it doesn't regress when we regenerate afmtodit.tables.

	* src/utils/afmtodit/make-afmtodit-tables: Do it.
	* src/utils/afmtodit/afmtodit.tables: Regenerate it.

2020-09-02  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/groff.1.man (Options) <-E>:
	* src/roff/troff/troff.1.man (Options) <-E>: Document that -E
	implies -Ww.

	Fixes <https://savannah.gnu.org/bugs/index.php?59030>.

2020-09-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff_char.7.man: Revise glyph descriptions.

	Fixes <https://savannah.gnu.org/bugs/index.php?59031>.

2020-09-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tty.tmac: Define fallback glyphs for Bell Systems logo
	{non-breaking adjustable space} and radical extension and square
	root extension (both non-printing input breaks) to suppress
	warnings from groff_char(7).  As none of these are encoded in
	Unicode it seems unlikely they will become supported soon.

2020-08-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/troff.1.man (Options) <-I>: The current working
	directory is searched _last_, not first, unless the order is
	altered with "-I .".

	* src/roff/groff/groff.1.man (Options) <-I>: Rearrange
	description.

2020-08-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	Update mdoc package to honor HY register as man does now, per
	suggestion from Colin Watson.  Recognize but ignore AD string
	for man package compatibility (essentially "reserving" it),
	though mdoc intentionally does not permit configuration of
	adjustment.

	* tmac/mdoc/doc-ditroff-u:
	* tmac/mdoc/doc-nroff-u: Do it.

	* tmac/mdoc/doc-common-u: Initialize \n[doc-hyphen-flags] from
	\n[HY].

	* tmac/groff_mdoc.7.man \
	(FORMATTING WITH GROFF, TROFF, AND NROFF): Document it.

2020-08-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (TH): Reset inter-word and inter-sentence
	spacing to default with each new page rendered.

	{To observe the problem prior to this change,
	  $ groff -mandoc groff_mmse.7 $any_other_man_page
	groff_mmse(7) loads sv.tmac, which changes the sentence
	spacing.}

2020-08-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* PROBLEMS: Undocument the problem with test-groff and mdoc.
	It's resolved.

	Fixes <https://savannah.gnu.org/bugs/index.php?51073>.

2020-08-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tmac.am (NORMALFILES, MDOCFILES): Drop unused variables.
	($(TMACMDOCFILES) $(TMACSTRIPFILES)): Scrub trailing whitespace.
	Wrap long line.

2020-08-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	mdoc: Relocate within build tree.

	The (modern) mdoc macro package has not been usable within the
	build tree, unlike the others.  This makes it more troublesome
	to test changes, and frustrates deploying our test
	infrastructure against it.  Re-arrange the build tree to
	resemble an installation tree closely enough for the macro
	package to load.

	* tmac/doc-common-u:
	* tmac/doc-ditroff-u:
	* tmac/doc-nroff-u:
	* tmac/doc-syms-u: Rename to...
	* tmac/mdoc/doc-common-u:
	* tmac/mdoc/doc-ditroff-u:
	* tmac/mdoc/doc-nroff-u:
	* tmac/mdoc/doc-syms-u: ...these.

	* tmac/tmac.am (TMACMDOCFILES): Look for files in tmac/mdoc.
	($(TMACMDOCFILES)): Create tmac/mdoc in build tree.

	Fixes <https://savannah.gnu.org/bugs/index.php?51003>.

2020-08-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test for usable in-tree mdoc.

	* tmac/tests/doc-smoke-test.sh: Add test.
	* tmac/tmac.am (tmac_TESTS): Run test.

2020-08-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tests/an-old_TH_repairs_ad_damage.sh:
	* tmac/tests/an-old_TH_repairs_hy_damage.sh: Add tests.
	* tmac/tmac.am (tmac_TESTS): Run tests.

2020-08-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (TH): Set hyphenation mode to user preference
	with each new page rendered, in case a hostile page meddled with
	'.hy' or '.nh'.
	(after .mso man.local): Relocate setting of default hyphenation
	mode here (instead of just before .mso man.local).  Eliminate
	conditional on \n[an-html]; this doesn't matter because
	grohtml(1) doesn't support hyphenation anyway.  If it does learn
	to hyphenate, we should treat it the same as other output
	devices in any case.  Add comment explaining why we fall back to
	different defaults depending on \n[cR].

2020-08-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (TH): Set adjustment to user preference with
	each new page rendered, in case a hostile page meddled with
	'.ad' or '.na'.
	(after .mso man.local): Set the default adjustment mode only if
	a -d option or man.local did not.  Also do the same for the HF
	string; relocate its definition here and make it similarly
	conditional.  Also update comment on setting of CS and CT
	registers and style their conditionals consistently with the
	rest of the package.

	* tmac/groff_man.7.man.in (Strings) <\*(AD>: Document it.
	(Strings) <\*(HF>: Parallelize language with new \*(AD.
	(Options) Expand introductory sentence to discuss -d and -r
	options.
	(Options) <-dAD>: Document default and likely values.  Nobody
	right-justifies or centers man page text; send the curious to
	groff(7) for '.ad' request documentation.
	(Options) <-dHF>: Document default and steer people to groff(7)
	for '.ft' request documentation.

2020-08-21  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac (set-an-margin): Rename...
	(an-set-margin): ...to this.  It was the only package-internal
	register, macro, string, or diversion that wasn't named using
	the "an-*" schema.
	(TH, SH, SS): Update call sites.

2020-08-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-ext.tmac (.ME, .UE): Restore hyphenation after
	"punctuation" arguments have been output.  The next token will
	always be a space node, the end of the document, or similar.

	* tmac/tmac.am (tmac_XFAIL_TESTS): Mark the punctuation
	hyphenation tests as expected to fail, because they now do.

2020-08-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/ps.tmac: Define ordinary hyphen-minus as fallback
	character for U+2011 (non-breaking hyphen).  Prompted by
	discussion with Dave Kemper in Savannah #58390.

2020-08-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	Fix hyphenation bug.  The UR/UE and MT/ME macros were much too
	aggressive about turning hyphenation off.  Disable it only when
	writing the actual URL or email address.

	* tmac/an-ext.tmac (.MT, .UR): Stop disabling hyphenation here.
	(.ME, .UE): Disable hyphenation right before output of
	URL/address; restore it right before output of supplementary
	arguments ("punctuation").

	* tmac/tests/an-ext_ME_punct_hyphenates.sh:
	* tmac/tests/an-ext_MT_body_hyphenates.sh:
	* tmac/tests/an-ext_UE_punct_hyphenates.sh:
	* tmac/tests/an-ext_UR_body_hyphenates.sh: Test behavior.

	* tmac/tmac.am (tmac_TESTS): Run tests.

2020-08-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	Adapt to the fact that \: reënables hyphenation.  Rename
	"hyphenless [discretionary] break" to "non-printing break point"
	in documentation.

	* doc/groff.texi (Manipulating Hyphenation): Rename "hyphenless
	break" to "non-printing break point".  Clarify that it is the
	soft hyphen glyph, not necessarily a hyphen, that is ordinarily
	written to the output on hyphenation breaks.  Note that the
	remainder of a word after \: is subject to hyphenation as
	normal.  Note (new) '\:\%' idiom for getting what people will
	want at least some of the time.  Update example to use it.
	* makevarescape.sed: Insert \% after we insert \: to rewrite
	slashes in filenames, to protect later portions of the filename
	from hyphenation.
	* man/groff.7.man (Escape Sequences/Escape short reference):
	Rename "hyphenless break" to "non-printing break point".
	* man/groff_diff.7.man (Language/Escape sequences): Resync with
	our Texinfo manual.
	* tmac/groff_man.7.man.in (Description/Hyperlink and email
	macros): Rename "hyphenless break" to "non-printing break
	point".

2020-08-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	documentation: Re-christen 'ESCAPE_AMPERSAND'.

	s/zero[- ]width space character/non-printing input break/

	* doc/groff.texi (Requests): Rename.  Update conceptual index
	entries; retain old name (with an appended "[sic]") to aid
	readers accustomed to it.
	(Ligatures and kerning): Update conceptual index entries.
	Supply context ("effect on kerning").
	(Drawing requests): Update conceptual index entries.  Supply
	context ("effect on '\l'").
	* man/groff.7.man (Description): Rename in macro-advice-writing
	shorthand.
	(Escape Sequences/Escape short reference): Rename.
	* tmac/groff_man.7.man.in (Description/Command synopsis macros
	[style]: Rename.
	(Description/Portability) [style]: Rename.
	* tmac/groff_mdoc.7.man (TROFF IDIOSYNCRASIES/Macro Usage):
	Rename.
	(TROFF IDIOSYNCRASIES/Other Possible Pitfalls): Rename.

2020-08-14  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_man.7.man.in (Description/{Document structure
	macros/.TH, Horizontal and vertical spacing}): Fix erroneous use
	of "flush left".  The current uses were introduced by me, but
	the page has borne incorrect uses of it as far back as 1999 (in
	the description of .TP).  Simply say "with no indentation"
	instead.

2020-08-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_man.7.man.in (Description/Number registers): Inform
	reader that registers can be set in man.local file and
	cross-reference it.
	(Files/*/man.local): [style] Supply example of man.local
	customizations, prompted by recent discussions on groff,
	linux-man, and man-db mailing lists.

2020-08-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tmac.am: Use a stamp file for m4 keyword check.
	(M4CHECK): Add stamp file variable.
	($(M4CHECK)): Create stamp file if check succeeds.

2020-08-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	Split groff_man(7) into two pages; one a (relatively) terse
	reference and the other a tutorial and style guide.  Both are
	generated from the same source, which is processed by m4 into
	two man page sources.

	* tmac/groff_man.7.man: Rename...
	* tmac/groff_man.7.man.in: ...to this.

	* tmac/groff_man.7.man.in: Add m4 directives and define macros.
	Protect m4 keywords in English from unintended expansion (this
	affected the word "include").  Add new .TH and apropos lines for
	style guide.  Convert marker comments into m4 macros for content
	control.  Uncomment material intended only for basic reference
	page.

	* tmac/tmac.am (man7_MANS): Add groff_man_style.7.
	(EXTRADIST) Add groff_man.7.man.in.
	(MOSTLYCLEANFILES): Add m4-generated man page sources
	tmac/groff_man{,_style}.7.man.
	Add targets to generate those same two pages from the renamed
	file.  Add target to grep the page for unprotected English m4
	keywords and halt the build if they are found.  Make generation
	of those pages depend on this new target.

2020-08-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_man.7.man: Add material on hooks and encoding.
	(Description, ./Deprecated features): Stop identifying .BT and
	.DT as "deprecated"; they shouldn't ever be called in man pages,
	but that has never been their intention.
	(Description): Identify character encoding and line-ending
	requirements (tutorial/style-guide material).
	(Description/Hooks): Add new subsection to house descriptions of
	.BT and .DT.  Add mnemonic expansions for both.

2020-07-31  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff.7.man (Control Characters): Incorrect claims were
	made.  Double quotes can indeed be used to enclose arguments to
	string interpolations.  The statement about a leading " in a
	string definition (or appendment with .as) was stated too
	generally.  The leading quote is not necessary to include
	leading tab characters, which can be input as-is, even in
	compatibility mode.

2020-07-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff.7.man (Requests/Request short reference/.ft): Fix
	some outright damage in the description of the request; the
	escape sequences that are synonymous with a nullary .ft had been
	wrongly removed.

2020-07-25  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/meintro.me:
	* doc/meref.me:
	* tmac/e.tmac-u: Remove postal address for Eric Allman.  It's
	probably decades out of date.  Adjust footnote symbols.  Thanks
	to Dave Kemper for the bug report and the patch.  Fixes
	<https://savannah.gnu.org/bugs/?58726>.

2020-07-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	Migrate macro packages from font CW to CR.  The font name "CW"
	{"constant-width"} is a legacy item we can dispose of (except
	for the DVI device, which also has CWI for "constant-width
	italic"), and much of the groff codebase already did starting
	with commit 0de1d6d79cdb959ffa7dac3af77c2a36ef31873f {October
	2017}.  Courier is available in the usual four styles {roman,
	bold, italic, bold+italic}.

	* tmac/an-ext.tmac (.EX): Set font to R, not CW.  The existing
	.ft request was somewhat redundant with the '.do fam C'
	immediately before, which set the font family to Courier.
	* tmac/s.tmac (.UC): Use font CR, not CW.

2020-07-23  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tty.tmac: Stop suppressing nonexistent font warning.
	Prompted by a 2017 suggestion and patch from Bjarni Ingi
	Gislason.  Documents and macro packages that want to change the
	font family should do so in awareness that this is meaningless
	on terminal (nroff) devices (and -Thtml as well).

	Fixes <https://savannah.gnu.org/bugs/?51364>.

2020-07-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_man.7.man (.PP, .LP, .P): Fix error: these macros
	do not reset the left margin.

2020-07-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Strings): Document behavior of .ds request
	when only one argument is supplied.  (The string is defined as
	empty.)  Thanks to Dave Kemper for the report and the patch.

	Fixes <https://savannah.gnu.org/bugs/index.php?58746>.

2020-07-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Using Symbols): Fix error in .rfschar
	description, which should refer to "font f" rather than "glyph
	f".  Thanks to Dave Kemper for the report and the patch.

	Fixes <https://savannah.gnu.org/bugs/index.php?58767>.

2020-07-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_man.7.man (.EE, .YS): Fix error: refer to "initial
	hyphenation setting" instead of "previous hyphenation mode".
	Incidentally, this fact points out why putting ".hy 0" or ".nh"
	in your man page is futile.

2020-06-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff.7.man (Requests/Request short reference): Fix error
	in description of .hc; it changes the hyphenation character
	rather than supplying an additional one.

2020-06-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Groff Options): Remove editorial comment about
	'-a' option being "useless".  It isn't.  Update example for
	contemporary systems (like Debian) and to reflect the fact that
	the GNU troff(1) man page needs to be preprocessed with tbl(1).
	(Invoking grotty): Recast discussion of -c option, importing
	much language from grotty(1) page rewrite from a year ago.  Add
	program index entries for col, more, and ul.  Fix transposition
	error in ISO document number.

	* src/devices/grotty/grotty.1.man (Description/Legacy output
	format): Make slight wording changes prompted by content of
	parallel section in our Texinfo manual.

	* src/roff/groff/groff.1.man (Options/-a): Parallelize with
	first sentence of corresponding material in our Texinfo
	manual.

	* src/roff/troff/troff.1.man (Options/-a): Parallelize with
	our Texinfo manual.

	Fixes the rest of
	<https://savannah.gnu.org/bugs/index.php?55278>.

2020-06-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff_diff.7.man (Language/Long names): Fix error: groff
	adds three new requests with short names, not two.  List them in
	an adjacent comment.

2020-06-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-ext.tmac (.ME, .UE): Fix portability issue.  While
	widely supported, the \$* escape is not documented in CSTR #54.
	If groff is not the typesetter, append macro arguments using
	\$1, \$2, ..., \$9 instead.

2020-05-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* man/groff.7.man (Numerical Expressions): Fix error: negative
	expressions evaluate false, not true as was implied.
	Parallelize descriptions with our Texinfo manual and
	groff_diff(7) (in abbreviated form).

2020-05-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Manipulating Filling and Adjusting): Fix error
	in code sample: missing 'n' in number register interpolation.

2020-05-14  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/preconv/tests/smoke-test.sh: Make BOM detection
	override less hinky (i.e., use more normative input).  Add tests
	for all five detected BOMs.

2020-05-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/env.cpp (space_size): Prevent assertion
	failure.  If an argument to the .ss request is negative, throw a
	range warning and ignore it.
	* src/roff/groff/tests/regression_savannah_58337.sh: Add test.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

	Fixes https://savannah.gnu.org/bugs/index.php?58337.

2020-05-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	Update documentation of .ss request.

	* doc/groff.texi (Manipulating Filling and Adjusting): Rewrite
	documentation of .ss request and \n[.ss] and \n[.sss] escapes.
	Note that negative values are not permitted in either argument
	to .ss.  Use new terminology, "minimal inter-word spacing" and
	"additional inter-sentence spacing" for clarity.  Clarify that
	additional inter-sentence spacing is only used when the output
	line is not full when the end of a sentence occurs.  Add index
	nodes to help readers find discussion of details of spacing
	between words and sentences.  Move discussion of differences
	from AT&T troff to section "Implementation Differences".  Update
	example to more closely resemble a real-world case, use second
	argument to .ss request, and eliminate hackish use of .nop
	request.

	* man/groff_diff.7.man (New number registers): Recast in
	parallel with the foregoing.

	* man/groff.7.man (Read-only registers): Use new terminology and
	describe \n[.ss] and \n[.sss] in meaningful terms, not by
	reference to arguments to the .ss request (which wasn't even
	accurate, because these registers have default values).

	Fixes https://savannah.gnu.org/bugs/index.php?54101.

2020-05-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	Undocument plans to support end-of-file GNU Emacs coding tags.

	* src/preproc/preconv/preconv.cpp (check_coding_tag):  Update
	comments.
	(detect_file_encoding): Alter debugging output so it's easier to
	grep and verify Emacs coding tag detection.

	* src/preproc/preconv/preconv.1.man (Bugs): Delete; its sole
	concern was the absence of this feature.
	(Usage): Document detection algorithm in more detail.  Note
	which detection methods don't work on unseekable input (pipes).
	Offer recommendations for those struggling with encoding
	detection.
	(Usage/Coding Tags): Stop manipulating line adjustment.  Use
	hyphen-minus (\- escape) characters in coding tag names, since
	they are literals that one might copy and paste.  Stop
	referencing XEmacs, whose development appears moribund.
	(See Also): Add cross-references to iconv(3) and locale(7) man
	pages.

	* src/preproc/preconv/tests/smoke-test.sh: Test each of the
	steps in the detection algorithm.
	* src/preproc/preconv/preconv.am: Run test.
	(preconv_TESTS): Add new variable.
	(TESTS): Append $(preconv_TESTS).

2020-05-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/afmtodit/afmtodit.pl: Format usage message with
	full capitals for parameters, and break out -v option in a
	separate "synopsis".

2020-05-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	* font/devpdf/util/BuildFoundries.pl: Stop throwing away
	diagnostics from afmtodit.  This results in only one extra line
	of build output:

		both Upsilon1 and Upsilon map to *U at .../afmtodit line
		6413.

2020-05-05  G. Branden Robinson <g.branden.robinson@gmail.com>

	Correct documentation of .pm request.

	* man/groff.7.man (Requests/Request short reference): The .pm
	request's classical behavior was described instead of groff's,
	and omitted mention of strings and diversions; correct it.

	* man/groff_diff.7.man (Implementation Differences): Document
	this difference between AT&T troff and groff.

	* doc/groff.texi (Debugging): Relocate description of .pm
	behavioral difference...
	(Implementation Differences): ...hither.

2020-04-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/addftinfo/addftinfo.cpp (usage):
	* src/utils/tfmtodit/tfmtodit.cpp (usage): Add "usage:" prefix
	to messages documenting auxiliary modes of invoking the program.
	The output doesn't look right without one, and it feels
	dishonest to not document the relevant options (-v, --version)
	disjunctively.  I'm trying to strike a balance between the
	ultra-terse BSD approach and the ultra-garrulous GNU one (see,
	e.g., ls(1)).  Likely both camps will be unhappy.  :-/

2020-04-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/tfmtodit/tfmtodit.cpp (read_map): Report invalid
	character code from map file in diagnostic.
	(main): Report invalid skew character position in diagnostic.
	Explicitly report insufficient arguments in addition to usage
	message.  Lift invariant computed expressions out of loops (more
	to shorten long lines than because I think the compiler won't
	figure it out).  Give content to "impossible assertion".
	(tfm::load): Capitalize TFM initialism in diagnostics.
	(usage): Use full capitals for parameters as is conventional.
	Document disjunction of -v/--version flag from other
	invocations.

2020-04-29  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/utils/addftinfo/addftinfo.cpp (usage): Add overloaded
	version that accepts a constant string argument, which emits the
	argument as a diagnostic and then calls usage().
	(main): Add diagnostics to usage message where the problem is
	clear.  Add comment explaining why it sometimes isn't.
	(usage): Refactor main usage message (which prints the summary)
	to report the actual names of the accepted option arguments
	instead of just "-param", which is not literally accepted.

	Sort param_table alphabetically for use by the usage message.

2020-04-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	Delete groffer.

	Per discussion on the groff development mailing list, there is
	no desire to retain the maintenance of this portion of the groff
	distribution.

	See
	<https://lists.gnu.org/archive/html/groff/2020-04/msg00051.html>
	et seq.

	* contrib/groffer: Recursively delete.

	* Makefile.am:
	* arch/mingw/mingw.am: Stop building groffer.

	* m4/groff.m4 (GROFF_GROFFERDIR_DEFAULT):
	(GROFF_GROFFERDIR_OPTION): Delete; remove "--with-grofferdir"
	configuration option.
	* configure.ac: Stop calling these macros.

	* MANIFEST:
	* NEWS:
	* PROJECTS:
	* TODO:
	* contrib/chem/chem.1.man:
	* contrib/chem/examples/122/README.txt:
	* contrib/chem/examples/README.txt:
	* contrib/glilypond/glilypond.1.man:
	* contrib/gperl/gperl.1.man:
	* contrib/gpinyin/gpinyin.1.man:
	* contrib/groff_filenames/groff_filenames.5.man:
	* man/groff_font.5.man:
	* man/roff.7.man:
	* src/roff/groff/groff.1.man:
	* src/roff/grog/grog.1.man:
	* src/utils/addftinfo/addftinfo.1.man:
	* tmac/groff_trace.7.man: Delete references to groffer.

2020-04-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/include/curtime.h: #include <time.h> if we're returning
	a time_t from current_time().

	* src/libs/libgroff/curtime.cpp (current_time): Quote
	$SOURCE_DATE_EPOCH variable content in diagnostics produced due
	to bad input since it's under user control and could have all
	kinds of bogosity in it (like whitespace).

2020-04-19  G. Branden Robinson <g.branden.robinson@gmail.com>

	* **/*.{man,tmac}: Save compatibility mode robustly.

	Use new \n[.cp] register to save compatibilty mode.

	Use register names based on the filename (at the source
	maintenance level) to avoid clobbering other files' saved
	compatibility modes.

	* tmac/html.tmac: Eliminate test of saved-compatibility
	register by moving its body inside the block where compatibility
	mode is off.  This is the only part of this changeset that was
	not automated.

	Fixes <https://savannah.gnu.org/bugs/index.php?58162>.

2020-04-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	Implement new read-only .cp register.

	Within a .do request, \n[.cp] holds the saved value of
	compatibility mode.

	The register \n[.cp] is specialized and may require a statement
	of rationale.  When writing macro packages or documents that use
	groff features and which may be mixed with other packages or
	documents that do not—common scenarios include serial processing
	of man pages or use of the .so or .mso requests—you may desire
	correct operation regardless of compatibility mode in the
	surrounding context.  It may occur to you to save the existing
	value of \n(.C into a register, say, _C, at the beginning of
	your file, turn compatibility mode off with “.cp 0”, then
	restore it from that register at the end with “.cp \n(_C”.  At
	the same time, a modular design of a document or macro package
	may lead you to multiple layers of inclusion.  You cannot use
	the same register name everywhere or you risk “clobbering” the
	value from a preceding or enclosing context.  The two‐character
	register namespace of AT&T troff is confining and mnemonically
	challenging; you may wish to use groff's more capacious
	namespace.  However, attempting “.nr _my_saved_C \n(.C” will not
	work in compatibility mode; the register name is too long.
	“This is exactly what .do is for,” you think, “.do nr
	_my_saved_C \n(.C”.  The foregoing will always save zero to your
	register, because .do turns compatibility mode off while it
	interprets its argument list.  What you need is:
		.do nr _my_saved_C \n[.cp]
		.cp 0
	at the beginning of your file, followed by
		.cp \n[_my_saved_C]
	at the end.  As in the C language, we all have to share one big
	namespace, so choose a register name that is unlikely to collide
	with other uses.

	* src/roff/troff/input.cpp (do_request, init_input_requests):
	Implement it.

	* doc/groff.texi:
	* man/groff.7.man:
	* man/groff_diff.7.man: Document it.

	* src/roff/groff/tests/dot-cp_register_works.sh: Test it.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

	Enables a fix for
	<https://savannah.gnu.org/bugs/index.php?58162>.

	Thanks to John Gardner and Ingo Schwarze for the discussion.

2020-04-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Implementation Differences):
	* man/groff_diff.7.man (New requests): Rewrite description of
	.do request, and replace example with a more illustrative one.

	* man/groff.7.man (Request short reference): Rewrite description
	of .do request briefly.

2020-04-14  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Implementation Differences): Remove incorrect
	claim.  The .C register cannot be manipulated manually.

2020-04-14  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/tests/smoke-test_html_device.sh: Simplify
	charmap test.

2020-04-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	It's too easy for the nroff version to get desynced from the
	groff version when we're running test cases, leading to spurious
	results.  Make it easier to see a discrepancy.

	* src/roff/nroff/nroff.sh: Call groff with -v or --version when
	we are called that way.  Let test cases tell us to use
	test-groff as groff with an environment variable.

	* src/roff/nroff/tests/verbose_option_works.sh.in: Rename to...
	* src/roff/nroff/tests/verbose_option_works.sh: ...this.  Check
	the nroff version being tested against the groff version being
	wrapped.  This exposes a bug; the system groff rather than the
	build tree groff was being invoked.  Refactor.  Stop messing
	with @GROFF_BIN_PATH_SETUP@ (enabling the rename); instead, let
	test-groff handle that for us.  Locate it and export the
	variable GROFF_TEST_GROFF so nroff can find it.  Because we
	manipulate $PATH to run the tests, the $PATH of an installed
	groff system will _always_ differ from that used by a build
	tree; ignore it in the test cases.  Dispose of bashisms and set
	shebang to /bin/sh.

	* src/roff/nroff/nroff.am: Stop generating the above test
	script.  It can now be run as it ships.

2020-04-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	Make our assert() C99-conformant.

	groff has its own implementation of the standard C library's
	assert() macro.  It hasn't been updated since C89.  C99 requires
	that the diagnostic emitted by assert() contain the failing
	expression and name of the function in scope.

	* src/include/assert.h: Add additional pointer to const char
	arguments to do_assert() and assertion_failed() for function
	name and stringified expression.
	(assertion_failed): Update prototype.
	(do_assert): Accept 'func' and 'msg' parameters and pass them to
	assertion_failed().
	(assert): Update macro to collect '__func__' and stringify the
	expr parameter (as 'msg') and pass them do do_assert().
	* src/libs/libgroff/assert.cpp (assertion_failed): Rewrite
	diagnostic to more closely match GNU Coding Style format and
	also report function and failing expression.

	Example output:
		troff: ../src/roff/troff/input.cpp:2644: do_request():
		assertion failed: '0 == "But first, here's a rotten old
		BBC programme."'

	We have no excuse to assert(0) ever again.  Express the
	invariant that has been violated.

2020-04-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/tests/smoke-test_html_device.sh: Set
	LC_CTYPE=C.UTF-8 so that byte sequences in the pipelines are
	handled correctly.  Skip the test if the environment doesn't
	support UTF-8.  Thanks to Bjarni Ingi Gislason for the trouble
	report.

	Fixes <https://savannah.gnu.org/bugs/?58164>.

2020-04-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/html.tmac: Use .do so we correctly load unicode.tmac in
	compatibility mode.

	Fixes the following problem (wrapped, filenames abbreviated):

	$ echo | ./build/test-groff -C -Thtml >/dev/null
	troff: backtrace: file '.../groff/build/../tmac/html.tmac':546
	troff: backtrace: file '.../groff/build/../tmac/troffrc':30
	troff: .../groff/build/../tmac/html.tmac:546: warning: macro
		'ms' not defined

2020-04-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/tests/smoke-test_html_device.sh: Add test.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

2020-04-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/tests/regression_savannah_58153.sh:
	* src/roff/groff/tests/\
	use_point_size_escape_with_single_digit_arg.sh: Use "set -e" to
	ensure that multi-test script files don't hide problems.  There
	are arguments against "set -e" (and you can ask Greg Wooledge
	for all of them), but I'm so used to it from years of writing
	Debian package maintainer scripts that I feel comfortable with
	it.  It is less verbose than several alternatives (especially
	having one test script per case).

	* src/roff/groff/tests/string_case_xform_unicode_escape.sh:
	Update with respect to new failure output (in a comment only,
	since this is an XFAIL test).

	* tmac/tests/an-old_AT_and_UC_footer_saved_and_restored.sh:
	Drop unnecessary "|| exit 1" from end of script.  Incidentally,
	this script illustrates an alternative to "set -e".

2020-04-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	Rework documentation of .spreadwarn.

	* doc/groff.texi (Debugging):
	* man/groff.7.man (Request short reference):
	* man/groff_diff.7.man (New requests): Recast description of
	.spreadwarn request, some based on suggestions from Dave Kemper.
	Make more prominent the fact that spreading only applies to
	adjustment mode 'b'.  Document warning type used.

	Fixes half of Savannah #58035.

2020-04-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	Require Texinfo 5.0 (February 2013) at a minimum.

	* m4/groff.m4 (GROFF_MAKEINFO): Check for version 5.0 (increased
	from 4.8).  Update diagnostics and comments.

	* INSTALL.extra:
	* README:
	* doc/groff.texi:
	* doc/webpage.ms: Document updated requirement.

	* doc/fixinfo.sh: Delete; it is no longer necessary to work
	around Texinfo 4.x bugs.

	* doc/doc.am (EXTRA_DIST): Stop shipping fixinfo.sh.
	(.texi.html): Remove makeinfo < 5.0 version check and consequent
	execution of fixinfo.sh.  Update comment.

2020-04-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	Enable backtracing across process/file boundaries when errors or
	non-ignored warnings are encountered.

	Experimentation reveals that .so, .mso, and .pso requests acted
	as barriers to backtracing except when explicitly requested with
	the .backtrace request.  Judging by the git history, this
	behavior dates back to June 1991 or earlier.  The intention,
	according to a source comment, was only to suppress the
	backtrace output for the line corresponding to the outermost
	level of the input stack (commonly, a file argument to groff).
	Unfortunately, that wasn't its only effect.

	This change does result in one additional line of output for
	each error or (non-ignored) warning when -b is given.  However,
	I regard this as unobjectionable because {1} a backtrace was in
	fact explicitly requested; and {2} it seems a poor tradeoff to
	suppress most of the backtrace in some complicated and
	frustrating cases for the sake of one fewer line of backtrace
	output in a trivial one.

	Now, backtracing behaves the same no matter what triggers it.

	Fixes Savannah #58153.

	* src/roff/troff/input.cpp (file_iterator::backtrace): Call
	get_location() for its side effect of rewriting a filename of
	"-" to "<standard input>", for consistency with other diagnostic
	messages.  (In this class, this member function always returns
	1, so ignore the return value.  This fact is an essential part
	of what led to the bug; the conditional
		p && !p->get_location(0, &f, &n)
	which appeared in the for loop of input_stack::backtrace() prior
	to this change would always evaluate to false when a node of the
	file_iterator class was encountered.)
	(input_stack::backtrace): Replace member function body with that
	of input_stack::backtrace_all().
	(input_stack::backtrace_all): Delete.
	(backtrace_request): Update the only call site of the above.

	* src/roff/groff/tests/regression_savannah_58153.sh: Add test.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

	* src/roff/groff/tests/string_case_xform_errors.sh: Update
	regression test to not be confounded by additional line of
	backtrace output.  Remove bashism along the way.

	* doc/groff.texi:
	* man/groff.7.man:
	* man/groff_diff.7.man: Update documentation.

2020-04-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	Reorganize backtrace output.

	* src/roff/troff/input.cpp (string_iterator::backtrace):
	Prefix the output with the program name if it is known.  In a
	deviation from the GNU Coding Standards, put the string
	"backtrace:" immediately next instead of the file and line
	number; this is because, particularly in cases where the
	backtrace is most interesting or needed, it will produce
	multiple lines of output and the common prefix aids visual
	understanding.
	(file_iterator::backtrace): As above; also, change prefix for
	popened nodes in the iterator from "process" to "pipe".  This
	seemed preferable to omitting the information entirely or using
	an inscrutable sigil like 'f' or 'p'.  The advantage is that the
	output is better aligned--better ergonomics for those who have
	to cope with a lot of it.  Furthermore, output the filename only
	once.

	Example of old and new output:

	/tmp/inner:2: backtrace: file '/tmp/inner'
	echo .so /tmp/inner:1: backtrace: process 'echo .so /tmp/inner'
	/tmp/outer:1: backtrace: file '/tmp/outer'

	troff: backtrace: file '/tmp/inner':2
	troff: backtrace: pipe 'echo .so /tmp/inner':1
	troff: backtrace: file '/tmp/outer':1

2020-04-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (read_size): Add units to diagnostic.
	This function works only with device-specific basic units
	internally; users likely think in terms of point size, so add
	the basic unit suffix "u" to offer a hint about what's going on.

	Example:
	$ groff -ww >/dev/null
	\s[10]A\s[-12]B
	troff: <standard input>:1: warning: point-size escape results in
	non-positive size -2000u; set to 1u

	On a PostScript device with 1000 basic units to the point, the
	difference is important and would be confusing without the unit
	indication.

2020-04-08  G. Branden Robinson <g.branden.robinson@gmail.com>

	Correct and clarify point size documentation.

	Our Texinfo manual has long documented the request '.ps 0' as
	restoring the previous point size (just as '\s0' or '.ps' with
	no argument does), but this is incorrect; since groff 1.02 or
	earlier (June 1991), the request has not actually worked this
	way.  Instead, '.ps 0' sets the point size to 1 basic unit
	{though output drivers may clamp this to a higher value}.  This
	behavior is consistent with AT&T troff, Hierloom Doctools troff
	and, per Ingo Schwarze, Plan 9 troff.  (It is, however, not
	consistent with neatroff.)

	* doc/groff.texi (Changing Type Sizes): Stop claiming that '.ps
	0' works like '\s0'.  Note that the resulting (computed) point
	size, not the argument, is clamped.  Note that it is
	non-positive, not negative, point sizes that are clamped to 1u.
	Add (forward) cross-reference to section where \n[.ps] is
	defined since it is mentioned here.  Move explanation of special
	handling of zero argument to \s escape description.

	* man/groff.7.man (Request short reference): Explain .ps N
	independently of \s.  Note clamping behavior.

2020-04-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (read_size): Correctly brace 'else'
	block.

	* src/roff/groff/tests/\
	use_point_size_escape_with_single_digit_arg.sh:
	Check that we get a diagnostic when relying on ambiguous form.

2020-04-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp (read_size): Move special-case
	interpretation of the '\sN' form of point-size escapes when 'N'
	is 1, 2, or 3 to compatibility mode (groff -C) only, and throw
	error diagnostic with suggestion for remedy if encountered.

	Traditionally, '\s36A' is interpreted as "set point size to 36,
	then emit 'A'".  However, only values in the range 10-39 are
	handled specially; '\s40A' is interpreted as a four-point "0A".
	This is unlike anything else in *roff grammar; see \*, \$, \f,
	\F, \g, \k, \m, \M, \n, \V, \Y, and \z.

	To anticipate objections: Why not throw only a warning?  Because
	there isn't a warning category for supported but ambiguous
	syntax (this behavior of AT&T troff dates back to the mid-1970s
	but was not documented in the Troff User's Manual until 1992).
	Why not throw the error outside of compatibility mode too?
	Because outside of compatibility mode we (now) have an
	unambiguous parse.

	Background: The Graphic Systems C/A/T phototypesetter (the
	original device target for AT&T troff) only supported a few
	discrete point sizes in the range 6..36, so Ossanna
	special-cased the parser to do what the user must have meant.
	Kernighan warned of this in the 1992 revision of CSTR #54
	{§2.3}, and more recently, McIlroy referred to it as a "living
	fossil".

	See:
	  https://lists.gnu.org/archive/html/groff/2020-03/msg00054.html
	  https://lists.gnu.org/archive/html/groff/2020-04/msg00002.html
	  https://lists.gnu.org/archive/html/groff/2020-04/msg00015.html
	and follow-ups for discussion.

	* NEWS: Advise users of behavior change and offer guidance.

	* doc/groff.texi:
	* man/groff.7.man: Document the restriction of special handling
	of point-size arguments to '\s' to compatibility mode.

	* src/roff/groff/tests/\
	use_point_size_escape_with_single_digit_arg.sh: Add regression
	test.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

2020-04-04  G. Branden Robinson <g.branden.robinson@gmail.com>

	Improve point-size escape diagnostics.

	* src/roff/troff/input.cpp (read_size): Disclose context
	{point-size escape intepretation} in diagnostic messages.  When
	a "bad digit" is encountered, describe it if possible.  When a
	relative adjustment results in a negative point size, report the
	computed value.  Also rename a variable for slightly more
	clarity ("bad" does not refer to all bad parses, just some cases
	of bad digits), and update an insufficiently generalized comment
	{"\s(00" is also an acceptable expression for point-size zero}.

	Based on suggestions by Ingo Schwarze and Bjarni Ingi Gislason.

2020-04-01  G. Branden Robinson <g.branden.robinson@gmail.com>

	Align diagnostic message format.

	* src/libs/libgroff/error.cpp (do_error_with_file_and_line):
	* src/roff/troff/input.cpp (do_error):
	Display the diagnostic level when it is "error", instead of
	leaving it implicit.

	* src/roff/troff/input.cpp (error_with_file_and_line): Include
	the program name in the diagnostic message if it is known.

	* src/roff/groff/tests/string_case_xform_errors.sh: Update
	diagnostic message expectation.

2020-02-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi (Page Motions):
	* man/groff.7.man (Single-character escapes): Document
	non-breaking spaces as such.

	Report and patch by Dave Kemper <saint.snit@gmail.com>.
	Fixes <https://savannah.gnu.org/bugs/index.php?55954>.

2020-02-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	nroff: Implement -P and -V options.

	* src/roff/nroff/nroff.sh: Recognize -P option (with argument)
	and pass it through to groff.  Recognize -V option; if given,
	display constructed groff command instead of executing it.
	Update usage message to document new options.  Tidy up shell
	style and indentation.  Add comments.
	* src/roff/nroff/nroff.1.man: Document new -P and -V options.
	* src/roff/nroff/tests/verbose_option_works.sh.in: Test -V
	option.
	* src/roff/nroff/nroff.am: Generate test and run it.

2020-02-06  G. Branden Robinson <g.branden.robinson@gmail.com>

	Stop ms macro package from manipulating warnings.

	* tmac/s.tmac: Remove code block that attempted to enable all
	warnings "only if none are given on the command line".  It did
	this by simply comparing the value of the warn register
	{\n[.warn]} against the default value; but of course, a user
	could specify -w options that exactly matched the default and
	the test would not be able to tell, causing puzzling and
	undesired behavior.  Furthermore, the hard-coded default was out
	of date and did not correspond to recent releases of groff.  If
	you want all warnings on, use the ".warn" request with no
	arguments in your ms document or pass "-w w" to groff (see
	troff(1) or our Texinfo manual for more on warnings).

	Thanks to Bjarni Ingi Gislason for bringing this issue to our
	attention.

	Resolves <https://savannah.gnu.org/bugs/?57583>.

2020-01-21  George HELFFRICH  <ghfbsd@gmail.com>

	[me]: Fix regression of '.nm' (#57638).

	* tmac/e.tmac-u (nm): Use saved point size and font position
	for line numbering only if those registers exist.

	Fixes <https://savannah.gnu.org/bugs/index.php?57638>.

2020-01-18  Ingo Schwarze  <schwarze@openbsd.org>

	Accept any number of arguments for .Dd in the groff_mdoc(7)
	macros.

	* tmac/doc-common-u: The .Dd macro behaved in a weird way:
	Without arguments, it printed the string "Epoch".
	With one, two, four, or more arguments, it ignored all arguments
	and used the current date instead.
	Only for exactly three arguments, it printed the arguments.
	None of this made sense.  Giving the date as "Epoch" is
	absurd, and printing the current date is just misleading:
	why should a document be considered up-to-date when the author
	did not even bother to state the date of the last change?
	Admittedly, the behaviour for 0 and 4 or more arguments
	already appeared 4.3BSD-Reno, and the behaviour for 2 or 3
	arguments in 4.4BSD.  But it was already wrong even in those
	days: several manual pages in 4.4BSD gave .Dd a single, quoted
	argument, e.g. .Dd "June 9, 1993": cap_mkdb(1), id(1), sed(1),
	err(3), getcap(3), sysctl(3), amd(8), disklabel(8), and others.
	Consequently, simply print all the arguments, no matter how
	many there are.

	This bug was found by Jonathan Gray <jsg@openbsd.org>
	while he looked at 4.xBSD manual pages.

2020-01-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* doc/groff.texi: Document initial empty set membership of
	characters with cflags values of 128, 256, or 512 in parallel
	with preceding paragraphs regarding smaller powers of two.

	Report and patch by Dave Kemper <saint.snit@gmail.com>.
	Fixes <https://savannah.gnu.org/bugs/?57549>.

2020-01-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/tests/*.sh: Rename some tests so their
	filenames are more self-descriptive.
	* src/roff/groff/groff.am (groff_TESTS): Update.

2020-01-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/groff/tests/transparent_end-of-sentence_chars.sh: Add
	test for end-of-sentence recognition.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

2020-01-18  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp: Mark \[dd] character (double dagger)
	as transparent for purposes of end-of-sentence recognition.

	* doc/groff.texi (Sentences):
	* man/groff_diff.7.man (New requests/.cflags): Document this.

	Report and patch by Dave Kemper <saint.snit@gmail.com>.
	Fixes <https://savannah.gnu.org/bugs/?57546>.

2020-01-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/refer/refer.1.man (See Also):
	* src/utils/indxbib/indxbib.1.man (See Also):
	* src/utils/lkbib/lkbib.1.man (See Also):
	* src/utils/lookbib/lookbib.1.man (See Also): Add bibliographic
	reference to the original Bell Labs "refer" paper by Mike Lesk.

2020-01-17  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tests/an-old_AT_and_UC_footer_saved_and_restored.sh: Add
	regression test for Ingo's change below.

2020-01-16  Ingo Schwarze  <schwarze@openbsd.org>

	Repair .AT and .UC in the groff_man(7) macros.

	* tmac/an-old.tmac: Setting user-defined strings in a macro that
	will later be called indirectly from page location traps is
	excessively complicated.  Besides, the implementation doesn't
	work: when the trap is finally sprung, the defaults from the
	an-init macro clobber what the author specified with .AT or .UC.
	Instead, all that is needed is setting the strings for the
	header before triggering the page break, such that they appear
	right away, while setting the strings for the footer after the
	page break, such that they don't appear on the previous page.

	This bug was found by Jonathan Gray <jsg@openbsd.org> while he
	looked at 4.xBSD manual pages.

2020-01-16  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/groff_man.7.man (Document structure macros/.SH): Fix
	errors in footer description.

2020-01-12  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/tests/an-old_CS_register_on.sh: Tweak flags; all the
	other tests in this group pass "-Tascii -P-cbou" instead of
	"-Tutf8".  (This is so that output is easy to grep from a shell
	script.)  Bring this test into line.  Does not regress anything.

2019-12-30  Deri James  <deri@chuzzlewit.myzen.co.uk>

	Certain pdfmark destination names caused gropdf to fail.

	* src/devices/gropdf/gropdf.pl: Look for pdfmark types, (i.e.
	DEST, OUT, ANN), only preceding 'pdfmark' at end of line, not
	anywhere else.

2019-12-30  Ingo Schwarze  <schwarze@openbsd.org>

	Correct output of sprintf("%%") in pic(1).

	* src/preproc/pic/pic.ypp: Print "%" rather than "%%".

	Bug reported by Doug McIlroy <doug at cs dot dartmouth dot edu>.
	Patch using feedback from Larry McVoy <lm at mcvoy dot com>.

2019-12-29  Deri James  <deri@chuzzlewit.myzen.co.uk>

	Update man page of gropdf to document \X calls.

	* src/devices/gropdf/gropdf.1.man: The calls 'pagename' and
	  'switchtopage' (used by mom to relocate TOC) are documented,
	  together with their convenience commands '.pdfpagename' and
	  '.pdfswitchtopage'.

2019-12-29  Ingo Schwarze  <schwarze@openbsd.org>

	Improve documentation of pic(1) regarding printf.

	* src/preproc/pic/pic.1.man:
	  Document which conversion specifications are supported.
	* doc/pic.ms:
	  Correct the list of supported conversion specifications.

2019-12-21  Ingo Schwarze  <schwarze@openbsd.org>

	Update NetBSD, OpenBSD, FreeBSD, Darwin, and DragonFly version
	strings.

	* tmac/doc-common-u: Update.
	* tmac/groff_mdoc.7.man: Synchronize.

	Based on a patch from Guillem Jover <guillem@debian.org>
	via Colin Watson <cjwatson@debian.org>, tweaked by me.
	Fixes: https://bugs.debian.org/867123

2019-09-22  G. Branden Robinson <g.branden.robinson@gmail.com>

	Use a vertical spacing of 1v between paragraphs in man page
	examples.  This looks better (on typesetter devices like
	PostScript and PDF) with the Courier font family also used in
	examples.  Typewriter devices (like the terminal) already used
	an inter-paragraph spacing of 1v.

	* tmac/an-ext.tmac (EX): Save the value of the PD number
	register (inter-paragraph distance), and set it to 1v.
	(EE): Restore previous value of the PD register.

2019-09-21  Deri James  <deri@chuzzlewit.myzen.co.uk>

	Changes to allow configure to check for URW fonts

	* font/devpdf/util/BuildFoundries.pl: Call the program with
	--dirURW with path provided to ./configure, and --check to do
	a dry-run just checking if the fonts are available. (bug #56748)

2019-09-21  Deri James  <deri@chuzzlewit.myzen.co.uk>

	Add new ghostscript font names (bug #56748)

	* font/devpdf/Foundry.in: Add changed font names

2019-09-21  Deri James  <deri@chuzzlewit.myzen.co.uk>

	Prevent gropdf executing arbitrary commands

	* src/devices/gropdf/gropdf.pl: See bug #55557

2019-09-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/an-old.tmac: Move test for definitions of CS and CT
	registers to after man.local is sourced; this way we can both
	respect any setting of those registers in that file (which is
	intended for customization by the site administrator) and ensure
	that the registers are defined when dereferenced later.

2019-09-15  G. Branden Robinson <g.branden.robinson@gmail.com>

	* **/*.man: Put section headings in title case.

2019-09-13  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/roff/troff/input.cpp: Lift invariant out of loop.

2019-09-11  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/preproc/refer/refer.1.man: Add mention of man and mm
	packages as supporting the "refer" preprocessor (GNU mm grew
	this support back in groff 1.22.1).

2019-09-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	Mark expected-to-fail test as XFAIL.

	Automake supports XFAIL and XPASS, so actually use the
	infrastructure as intended.

	* Makefile.am: Declare XFAIL_TESTS variable.
	* src/roff/groff/groff.am (groff_XFAIL_TESTS): Add
	string_case_xform_unicode_escape.sh.
	* src/roff/groff/tests/string_case_xform_unicode_escape.sh: Set
	the expected output to what it would be if string case
	transforms worked on groff Unicode escapes.

2019-09-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	Retire '@G@' build system macro.

	It existed only for forcibly-capitalized man page titles; it was
	the uppercase-transformed counterpart of @g@.

	* Makefile.am: Delete sed transformation of @G@.
	* src/preproc/tbl/tbl.1.man: Replace '@G@EQN' in section heading
	with '@g@eqn'.

2019-09-10  G. Branden Robinson <g.branden.robinson@gmail.com>

	* **/*.man: Downcase man page titles, except for glilypond and
	roff2*, which did not require it.

	Here's a sed script for the aid of those who have a corpus of
	man pages to migrate.

		# Downcase man page titles.
		#
		# Use the first line to skip any .TH lines that require
		# special handling, perhaps because the name of the
		# command gets externally macro-expanded.  In groff, the
		# "roff2.1.man" page is an example.
		#
		# This script operates _only_ on .TH lines.
		#
		# usage: sed -i -f this_script.sed your_man_page.1 ...
		/@ROFF2MODE@/b
		/^\.\s*TH/{
		h
		s/^\.\s*TH \+\([^ ]\+\) .*/\1/
		y/ABCDEFGHIJKLMNOPQRSTUVWXYZ/abcdefghijklmnopqrstuvwxyz/
		G
		s/\([^ ]\+\)\n\(^\.\s*TH \+\)[^ ]\+\( .*\)/\2\1\3/
		}

2019-09-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	The an (man) macro package supports new CS and CT number
	registers to control rendering of man page section headings and
	titles, respectively, in full capitals.  These default off (with
	no visible effect on pages which already fully capitalize these
	strings in man page sources).  The rationale is to encourage man
	page authors to preserve case distinction information in (or
	restore it to) their titles and section headings, while giving
	users (including system administrators, distributors,
	integrators, and maintainers of man(1) implementations) a way to
	view the rendered page elements in full capitals if desired.

	* tmac/an-old.tmac (TH/an-init): If number register CT is
	defined and nonzero, call .stringup on \*[an-title].
	(SH): Store macro argument list to \*[an-section-heading].  If
	number register CS is defined and nonzero, call .stringup on it.
	* tmac/groff_man.7.man (OPTIONS): Document the CS and CT number
	registers.
	* tmac/tests/an-old_CS_register_off.sh:
	* tmac/tests/an-old_CS_register_on.sh:
	* tmac/tests/an-old_CS_register_unspecified.sh:
	* tmac/tests/an-old_CT_register_off.sh:
	* tmac/tests/an-old_CT_register_on.sh:
	* tmac/tests/an-old_CT_register_unspecified.sh: Add tests.
	* tmac/tmac.am (tmac_TESTS): Run tests.

2019-09-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	Add regression test to check behavior of case transformation
	request on a string containing a Unicode character escape.
	Right now, this is not supported, so it is an XFAIL test in the
	parlance of DejaGNU/POSIX 1003.3.

	* src/roff/groff/tests/string_case_xform_unicode_escape.sh: New
	test.
	* src/roff/groff/groff.am (groff_TESTS): Run test.

2019-09-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	Implement .stringdown and .stringup requests.

	* src/roff/troff/input.cpp: Add .stringdown and .stringup
	requests.
	* doc/groff.texi: Document them, including example.
	* man/groff_diff.7.man: Same.
	* man/groff.7.man: Document them briefly.

2019-09-09  G. Branden Robinson <g.branden.robinson@gmail.com>

	Regression-test string case transform feature.

	* src/roff/groff/tests/string_case_xform_errors.sh: New test.
	* src/roff/groff/tests/string_case_xform_requests.sh: New test.
	* src/roff/groff/groff.am (groff_TESTS): Run tests.

2019-06-30  G. Branden Robinson <g.branden.robinson@gmail.com>

	* src/devices/grotty/grotty.1.man (Options/-i): Note support of
	recent versions of xterm for italic (oblique) styles.

	Fixes part of <https://savannah.gnu.org/bugs/index.php?55278>.

2019-06-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	devlatin1: Map \(oq to ' on output.

	* font/devlatin1/R.proto: Render the output glyph \[oq]
	{opening quote} as 0x27 (apostrophe) instead of 0x60 (grave
	accent).  A grave accent is just flat wrong for the Latin-1
	character set, a.k.a. ISO 8859-1.  The defining document ECMA-94
	{June 1986} is freely available for perusal.  The ECMA-94 Latin
	character sets do not define any glyphs for directional
	{"typographer's"} quotation marks, but the apostrophe is
	depicted as a neutral (vertical) glyph, whereas the grave accent
	0x60 and acute accent 0xB4 are mirror-symmetric diacritical
	marks.

	Note that this change has no effect on _input_ conventions for
	roff source documents.  You can still get directional single
	quotes on UTF-8, PostScript, PDF, and other output devices
	supporting them by typing sequences like `this' in the input
	{character remapping with ".char" requests and similar
	notwithstanding}.

	Patch and idea from Ingo Schwarze, who originally proposed it
	for ASCII as well, and included Latin-1 for parallelism.
	The groff developers could reach no consensus about the wisdom
	of such a change for ASCII (which was designed to support
	ambiguity for some code points, requiring the development of
	supplementary interpretation conventions between parties).
	ECMA-94/ISO-8859 is more strongly prescriptive.

	See https://savannah.gnu.org/bugs/?55616 and the groff mailing
	list archives for 31 January to 23 February 2019 at
	https://lists.gnu.org/archive/html/groff for lengthy discussion.

	* src/roff/groff/tests/on-latin1-device-oq-is-0x27.sh: Check for
	correct output glyph.

	* src/roff/groff/groff.am: Add test.

2019-06-28  G. Branden Robinson <g.branden.robinson@gmail.com>

	* tmac/unicode.tmac: Save and restore compatibility mode since
	the script uses a GNU extension (.char).

2019-06-27  G. Branden Robinson <g.branden.robinson@gmail.com>

	libdriver: Fix SEGV (Savannah #56555).

	* src/libs/libdriver/printer.cpp: Check result of
	set_char_and_width() for error condition before relying on it.

	* src/roff/groff/tests/regression-56555.sh: Provoke segfault
	with five bytes of input using transparent line indicator.

	* src/roff/groff/groff.am: Add test.

2019-01-12  Ingo Schwarze  <schwarze@openbsd.org>

	Correctly handle groff_mdoc(7) .Lk arguments starting with a
	dot.

	* tmac/doc.tmac-u: Each argument to the .Lk macro is printed
	on its own roff input line.  If it happens to start with a dot
	or apostroph, it was misinterpreted as a request or macro.
	Force it to be treated as text by adding "\&" in front of it.

2018-12-31  Ingo Schwarze  <schwarze@openbsd.org>

	* man/groff_diff.7.man: remove two stray .RE macros


________________________________________________________________________

##### License

Copyright 2018-2021 Free Software Foundation, Inc.

Copying and distribution of this file, with or without modification,
are permitted in any medium without royalty provided the copyright
notice and this notice are preserved.


##### Editor settings
Local Variables:
fill-column: 72
mode: change-log
version-control: never
End:
vim:set autoindent textwidth=72: