Add missing deprecation message for unescaped '{' in regexes

The use of literal '{' without being escaped has been deprecated since
5.16, and warned on since 5.20.  In 5.24, this has been made illegal,
with a bunch of CPAN modules broken by it, in spite of the long
deprecation period.  See
https://rt.perl.org/Ticket/Display.html?id=128139

Unfortunately, I overlooked a code path, and not all instances that
should have warned did so in fact.  This was spotted by Tom Wyant in
https://rt.perl.org/Ticket/Display.html?id=128213

This commit adds that warning, and rewords the fatal one slightly, and
clarifies the whole thing in perldiag.

4 files changed, 141 insertions, 19 deletions

diff --git a/pod/perldelta.pod b/pod/perldelta.pod
index 1997cfa5f0..fe27320058 100644
--- a/pod/perldelta.pod
+++ b/pod/perldelta.pod
@@ -271,7 +271,12 @@ Perl yourself.  The #! line at the top of your file could look like
 
 =item *
 
-XXX L<message|perldiag/"message">
+L<Unescaped left brace in regex is deprecated here, passed through in regex; marked by S<<-- HERE> in mE<sol>%sE<sol>|perldiag/"Unescaped left brace in regex is deprecated here, passed through in regex; marked by S<<-- HERE> in m/%s/">
+
+Unescaped left braces are already illegal in some contexts in regular
+expression patterns, but, due to an oversight, no deprecation warning
+was raised in other contexts where they are intended to become illegal.
+This warning is now raised in these contexts.
 
 =back
 
@@ -293,7 +298,11 @@ XXX Changes (i.e. rewording) of diagnostic messages go here
 
 =item *
 
-XXX Describe change here
+L<Unescaped left brace in regex is illegal here in regex; marked by S<<-- HERE> in mE<sol>%sE<sol>|perldiag/"Unescaped left brace in regex is illegal here in regex; marked by S<<-- HERE> in m/%s/">
+
+The word "here" has been added to the message that was raised in
+v5.25.1.  This is to indicate that there are contexts in which unescaped
+left braces are not (yet) illegal.
 
 =back
 
diff --git a/pod/perldiag.pod b/pod/perldiag.pod
index c0899ff6de..4164baf833 100644
--- a/pod/perldiag.pod
+++ b/pod/perldiag.pod
@@ -6139,20 +6139,117 @@ C<undef *foo>.
 (A) You've accidentally run your script through B<csh> instead of Perl.
 Check the #! line, or manually feed your script into Perl yourself.
 
-=item Unescaped left brace in regex is illegal in regex;
+=item Unescaped left brace in regex is deprecated here, passed through in
+regex; marked by S<<-- HERE> in m/%s/
+
+(D deprecated, regexp)  The simple rule to remember, if you want to
+match a literal C<"{"> character (U+007B C<LEFT CURLY BRACKET>) in a
+regular expression pattern, is to escape each literal instance of it in
+some way.  Generally easiest is to precede it with a backslash, like
+C<"\{"> or enclose it in square brackets (C<"[{]">).  If the pattern
+delimiters are also braces, any matching right brace (C<"}">) should
+also be escaped to avoid confusing the parser, for example,
+
+ qr{abc\{def\}ghi}
+
+Forcing literal C<"{"> characters to be escaped will enable the Perl
+language to be extended in various ways in future releases.  To avoid
+needlessly breaking existing code, the restriction is is not enforced in
+contexts where there are unlikely to ever be extensions that could
+conflict with the use there of C<"{"> as a literal.
+
+In this release of Perl, some literal uses of C<"{"> are fatal, and some
+still just deprecated.  This is because of an oversight:  some uses of a
+literal C<"{"> that should have raised a deprecation warning starting in
+v5.20 did not warn until v5.26.  By making the already-warned uses fatal
+now, some of the planned extensions can be made to the language sooner.
+
+The contexts where no warnings or errors are raised are:
+
+=over 4
+
+=item *
+
+as the first character in a pattern, or following C<"^"> indicating to
+anchor the match to the beginning of a line.
+
+=item *
+
+as the first character following a C<"|"> indicating alternation.
+
+=item *
+
+as the first character in a parenthesized grouping like
+
+ /foo({bar)/
+ /foo(?:{bar)/
+
+=item *
+
+as the first character following a quantifier
+
+ /\s*{/
+
+=back
+
+=for comment
+The text of the message above is duplicated below to allow splain (and
+'use diagnostics') to work.  Since one is fatal, and one not, they can't
+be combined as one message.  And since the non-fatal one is temporary,
+there's no real need to enhance perldiag to handle this transient case.
+
+=item Unescaped left brace in regex is illegal here in regex;
 marked by S<<-- HERE> in m/%s/
 
-(F) You used a literal C<"{"> character in a regular
-expression pattern.  You should change to use C<"\{"> or C<[{]> instead.
-If the pattern delimiters are also braces, any matching
-right brace (C<"}">) should also be escaped to avoid confusing the parser,
-for example,
+(F) The simple rule to remember, if you want to
+match a literal C<"{"> character (U+007B C<LEFT CURLY BRACKET>) in a
+regular expression pattern, is to escape each literal instance of it in
+some way.  Generally easiest is to precede it with a backslash, like
+C<"\{"> or enclose it in square brackets (C<"[{]">).  If the pattern
+delimiters are also braces, any matching right brace (C<"}">) should
+also be escaped to avoid confusing the parser, for example,
+
+ qr{abc\{def\}ghi}
+
+Forcing literal C<"{"> characters to be escaped will enable the Perl
+language to be extended in various ways in future releases.  To avoid
+needlessly breaking existing code, the restriction is is not enforced in
+contexts where there are unlikely to ever be extensions that could
+conflict with the use there of C<"{"> as a literal.
+
+In this release of Perl, some literal uses of C<"{"> are fatal, and some
+still just deprecated.  This is because of an oversight:  some uses of a
+literal C<"{"> that should have raised a deprecation warning starting in
+v5.20 did not warn until v5.26.  By making the already-warned uses fatal
+now, some of the planned extensions can be made to the language sooner.
+
+The contexts where no warnings or errors are raised are:
+
+=over 4
+
+=item *
 
-    qr{abc\{def\}ghi}
+as the first character in a pattern, or following C<"^"> indicating to
+anchor the match to the beginning of a line.
 
-This restriction is not enforced if the C<"{"> is the first character in
-the pattern; nor is a warning generated for this case, as there are no
-current plans to forbid it.
+=item *
+
+as the first character following a C<"|"> indicating alternation.
+
+=item *
+
+as the first character in a parenthesized grouping like
+
+ /foo({bar)/
+ /foo(?:{bar)/
+
+=item *
+
+as the first character following a quantifier
+
+ /\s*{/
+
+=back
 
 =item unexec of %s into %s failed!
 
diff --git a/regcomp.c b/regcomp.c
index 86173db3e0..2e7805768b 100644
--- a/regcomp.c
+++ b/regcomp.c
@@ -13259,7 +13259,7 @@ S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
 		     * something like "\b" */
 		    if (len || (p > RExC_start && isALPHA_A(*(p -1)))) {
                         RExC_parse = p + 1;
-			vFAIL("Unescaped left brace in regex is illegal");
+			vFAIL("Unescaped left brace in regex is illegal here");
 		    }
 		    /*FALLTHROUGH*/
 		default:    /* A literal character */
@@ -13664,8 +13664,6 @@ S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
 	    RExC_parse = p - 1;
             Set_Node_Cur_Length(ret, parse_start);
 	    RExC_parse = p;
-            skip_to_be_ignored_text(pRExC_state, &RExC_parse,
-                                    FALSE /* Don't force to /x */ );
 	    {
 		/* len is STRLEN which is unsigned, need to copy to signed */
 		IV iv = len;
@@ -13677,6 +13675,13 @@ S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
 	break;
     } /* End of giant switch on input character */
 
+    /* Position parse to next real character */
+    skip_to_be_ignored_text(pRExC_state, &RExC_parse,
+                                            FALSE /* Don't force to /x */ );
+    if (PASS2 && *RExC_parse == '{' && OP(ret) != SBOL && ! regcurly(RExC_parse)) {
+        ckWARNregdep(RExC_parse + 1, "Unescaped left brace in regex is deprecated here, passed through");
+    }
+
     return(ret);
 }
 
diff --git a/t/re/reg_mesg.t b/t/re/reg_mesg.t
index ff200834a3..e3c11ba4cf 100644
--- a/t/re/reg_mesg.t
+++ b/t/re/reg_mesg.t
@@ -268,10 +268,11 @@ my @death =
  '/(?[\ |!])/' => 'Incomplete expression within \'(?[ ])\' {#} m/(?[\ |!{#}])/',    # [perl #126180]
  '/(?[()-!])/' => 'Incomplete expression within \'(?[ ])\' {#} m/(?[()-!{#}])/',    # [perl #126204]
  '/(?[!()])/' => 'Incomplete expression within \'(?[ ])\' {#} m/(?[!(){#}])/',      # [perl #126404]
- '/\w{/' => 'Unescaped left brace in regex is illegal {#} m/\w{{#}/',
- '/\q{/' => 'Unescaped left brace in regex is illegal {#} m/\q{{#}/',
- '/:{4,a}/' => 'Unescaped left brace in regex is illegal {#} m/:{{#}4,a}/',
- '/xa{3\,4}y/' => 'Unescaped left brace in regex is illegal {#} m/xa{{#}3\,4}y/',
+ '/\w{/' => 'Unescaped left brace in regex is illegal here {#} m/\w{{#}/',
+ '/\q{/' => 'Unescaped left brace in regex is illegal here {#} m/\q{{#}/',
+ '/\A{/' => 'Unescaped left brace in regex is illegal here {#} m/\A{{#}/',
+ '/:{4,a}/' => 'Unescaped left brace in regex is illegal here {#} m/:{{#}4,a}/',
+ '/xa{3\,4}y/' => 'Unescaped left brace in regex is illegal here {#} m/xa{{#}3\,4}y/',
  '/abc/xix' => 'Only one /x regex modifier is allowed',
  '/(?xmsixp:abc)/' => 'Only one /x regex modifier is allowed {#} m/(?xmsixp{#}:abc)/',
  '/(?xmsixp)abc/' => 'Only one /x regex modifier is allowed {#} m/(?xmsixp{#})abc/',
@@ -621,6 +622,16 @@ my @experimental_regex_sets = (
 );
 
 my @deprecated = (
+ '/^{/'          => "",
+ '/foo|{/'       => "",
+ '/foo|^{/'      => "",
+ '/foo({bar)/'   => "",
+ '/foo(:?{bar)/' => "",
+ '/\s*{/'        => "",
+ '/a{3,4}{/'     => "",
+ '/.{/'         => 'Unescaped left brace in regex is deprecated here, passed through {#} m/.{{#}/',
+ '/[x]{/'       => 'Unescaped left brace in regex is deprecated here, passed through {#} m/[x]{{#}/',
+ '/\p{Latin}{/' => 'Unescaped left brace in regex is deprecated here, passed through {#} m/\p{Latin}{{#}/',
 );
 
 for my $strict ("", "use re 'strict';") {