| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
|
|
|
| |
We need to unref the cairo surface here, as we are making the context
be the sole owner of it.
https://bugzilla.gnome.org/show_bug.cgi?id=671687
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=671687
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=671687
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
It appears that cmph library can return (n+1) when querying item not
present in its original n-item-sized set. Adjust code so that it
detects this condition and do not chase stray pointers resulting from
this bogus(?) hash result.
https://bugzilla.gnome.org/show_bug.cgi?id=675939
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
We need to g_variant_ref() the return type: the closure delivers it wrapped
in a GValue, and the subsequent g_value_unset() unrefs it to zero otherwise.
This was uncovered by the now working TestClosures.test_variant() test case in
PyGObject.
Also allow passing NULL as argument, to be able to test these code paths as
well.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
C99 has standard types for fixed width signed and unsigned integers.
This adds types to giscanner to map them to the glib equivalent types.
stdint.h also has other types to specify specify sizes with at least a
certain number of bits, the fastest type with at least some number of
bits, a type big enough to store a pointer and a type containing the
most number of bits. These don't map well to glib types so I have
ignored them.
https://bugzilla.gnome.org/show_bug.cgi?id=674796
|
|
|
|
|
|
|
|
|
|
|
| |
Commit 7c4fbbd6 introduced some low-level g_memdup()ing of the passed GVariant
in regress_test_closure_variant(). This causes a race condition with GI clients
like pygobject's test suite, which sometimes fail with
(runtests.py:15653): GLib-CRITICAL **: g_variant_ref_sink: assertion `value->ref_count > 0' failed
Drop the memduping again, and instead make the argument non-const (which is
true, as the method temporarily increases the refcount).
|
|
|
|
| |
Should have signal arguments in the right order
|
| |
|
| |
|
|
|
|
|
|
|
| |
Add a flag and enum check to
the GHashes returned/expected by regress_test_ghash_gvalue_{return,in}().
https://bugzilla.gnome.org/show_bug.cgi?id=637466
|
|
|
|
|
|
|
|
|
|
|
| |
A test case with an array in-argument that is neither zero terminated,
nor fixed length, nor blessed with a length argument.
This is, for example, the case in clutter_texture_set_from_rgb_data()
https://bugzilla.gnome.org/show_bug.cgi?id=674271
Signed-off-by: Martin Pitt <martinpitt@gnome.org>
|
|
|
|
|
|
|
|
| |
Needed for https://bugzilla.gnome.org/show_bug.cgi?id=674351
https://bugzilla.gnome.org/show_bug.cgi?id=674366
Signed-off-by: Martin Pitt <martinpitt@gnome.org>
|
|
|
|
| |
Ran misc/update-glib-annotations.py against current glib master.
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=673806
|
|
|
|
|
|
|
|
|
| |
Including:
- handle things in the logical order encountered (first colon,
then annotations)
- correctly report column when missing a colon on the identifier part
- small type fixes
- remove no longer useful "parameter/tag expected" warnings
|
| |
|
|
|
|
|
|
|
|
|
| |
This reverts commit f41525d54f53edd9b0d665397f7c6c755ee3616b and
adds a more complete solution as the original commit only handled
empty lines between the GTK-Doc comment block start token "/**" and
the identifier.
https://bugzilla.gnome.org/show_bug.cgi?id=673806
|
|
|
|
| |
Assume srcdir=builddir, so users don't have to set this.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The current logic in _parse_comment_block of the annotationparser
check the identifier and use variables that only makes sense if a
result was returned (column_offset, original_line). This patch
indent the logic so it is only triggered if a result is found.
Otherwise it returns None as expected by call chain
"parse > parse_comment_block > _parse_comment_block":
ie in "parse":
comment_block = self.parse_comment_block(comment)
if comment_block is not None:
This still raises a lot of warnings but does not abort while generating
gir for libgnomekbd, mutter or telepathy-glib (those thus have to be
tweaked not to call g-ir-scanner with --warn-error at least until the
parser mask those false positives or the doc from those projects is
fixed to comply with the behaviour expected by giscanner).
https://bugzilla.gnome.org/show_bug.cgi?id=673806
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The intention of the change is to clean up unnecessary linking with
libffi and gmodule in g-i consumers as they typically don't use any
libffi/gmodule symbols.
Listing libffi and gmodule in pkg-config Requires meant that every
consumer that wanted to link with -lgirepository also got -lffi and
-lgmodule-2.0, which caused ld to emit unnecessary DT_NEEDED entries.
Moving the deps to Requires.private removes the internal dependencies
from 'pkg-config --libs', but keeps them in --libs --static output. The
change won't affect --cflags and the include directories necessary for
<ffi.h> will still be listed.
https://bugzilla.gnome.org/show_bug.cgi?id=674065
|
|
|
|
|
|
|
| |
Commit 2fff986 removed the use of g_thread_init() and we don't use any
other symbols from gthread-2.0.
https://bugzilla.gnome.org/show_bug.cgi?id=674065
|
|
|
|
|
| |
Address all gcc warnings, -Werror is not enabled yet but at least
-Wall -Werror passes on my machine.
|
|
|
|
|
|
| |
Makes it work on case-sensitive file systems such as HFS+.
https://bugzilla.gnome.org/show_bug.cgi?id=667405
|
|
|
|
|
|
|
|
| |
We should honor LDFLAGS environment variable set by the user,
so options such as -Wl,--enable-new-dtags properly get passed to
the linker when creating the dumper binary
https://bugzilla.gnome.org/show_bug.cgi?id=670152
|
|
|
|
|
|
|
|
| |
Most cases in validate() were already simply calls to _validate_option()
with particular arguments; this extracts the code from the remaining
options to their own methods, making the dispatch table more legible.
https://bugzilla.gnome.org/show_bug.cgi?id=670985
|
|
|
|
|
|
|
|
| |
"Maximium" is not an English word. Even if corrected to "maximum" (which
is), "at maximum" is not conventional usage: "at most" is more
idiomatic.
https://bugzilla.gnome.org/show_bug.cgi?id=670985
|
|
|
|
|
|
|
|
|
|
| |
Previously, the validation code compared each option name to a constant
(such as OPT_ALLOW_NONE) which expands to the string used in a source
file, and if they matched it would typically pass that same string to
_validate_option(). But the annotation name was written out longhand
each time, which seemed strange to me.
https://bugzilla.gnome.org/show_bug.cgi?id=670985
|
|
|
|
|
|
|
| |
This prevents a simple error from pkg-config, "Must specify package names
on command line"
https://bugzilla.gnome.org/show_bug.cgi?id=673668
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fix of leak in 4c9424e18d71237f438a99bc5f2d45ae7de60b78 was a bit
overaggressive, stealing also one reference from input 'info'
argument.
Also fixes another bug in that commit - local 'func' shadowing the
return value, causing that function always returned NULL even when
some result was actually found.
https://bugzilla.gnome.org/show_bug.cgi?id=673282
|
|
|
|
|
|
|
|
|
|
|
| |
This avoids tags overwriting parameters if they happen to
share the same name. For example, this was triggered by
valid code in libgnome-keyring.
tests/scanner/regress.c and tests/scanner/regress.h test
written by Colin Walters <walters@verbum.org>.
https://bugzilla.gnome.org/show_bug.cgi?id=672254
|
|
|
|
|
|
| |
The headers are already parsed as they are passed
in to g-ir-scanner. This avoids quite a few warnings triggered
with the new annotation parser.
|
|
|
|
|
| |
This removes also all the headers as it's the first run
with the updated extraction tool.
|
|
|
|
|
| |
This avoids a bit of python work and reduces the
amount of allocations.
|
|
|
|
|
| |
Only parse doc comments for the files we pass in
on the command line to g-ir-scanner, not the included dependencies.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- add missing colons
- invalid annotations
- invalid parameters and tags
- correct parameter name
- preserve description indentation
- no description parts
- comment end marker
- invalid empty lines
- line numbers
AnnotationParser now emits warnings which are considered as
errors by "make check" so fix those warnings...
https://bugzilla.gnome.org/show_bug.cgi?id=672254
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=672254
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=672254
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=672254
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=672254
|
|
|
|
|
|
|
|
|
|
|
| |
In Sandro Sigala's original c-c++-grammars code, this function
was known as "static void skip_comment(void)". It was built this
way instead of a lex pattern because it didn't need to save the
matched comment in the "yytext" buffer. This is still true as
we directly assign the matched comment to the scanner object.
So simply prepend and append the markers back where they belong.
https://bugzilla.gnome.org/show_bug.cgi?id=672254
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=672254
|
| |
|
|
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=663052
Signed-off-by: Martin Pitt <martinpitt@gnome.org>
|
|
|
|
|
|
|
|
| |
So it can be used for invoking callbacks
https://bugzilla.gnome.org/show_bug.cgi?id=663052
Signed-off-by: Martin Pitt <martinpitt@gnome.org>
|