Re: Bogus Bison 3.8.2 warning: suspicious sequence in the output

Hi Marko,

> Le 27 oct. 2021 à 15:04, Marko Mäkelä <[email protected]> a écrit :
> 
> Hi,
> 
> Bison 3.8.2 is issuing the following warning for a grammar whose semantic action includes the variable name my_charset_utf8mb4_bin:
> 
> yy_mariadb.cc:35834: warning: suspicious sequence in the output: b4_bin [-Wother]

Bummer...

> It seems to me that the regular expression in http://git.savannah.gnu.org/cgit/bison.git/tree/src/scan-skel.l is too greedy, or the check should be suppressed if the input contains such identifiers.

Let's start with "too greedy".  Collecting all the identifiers in the input is too much effort for this warning.

> A quick search of the Bison Texinfo for b4_ or m4_ did not find anything. Because those are not documented as reserved words or sequences, I think that identifiers like that should be allowed in semantic actions without Bison reporting any warnings.

The pattern was stupid and caught occurrences of b4_/m4_ anywhere.  The point was to rather to catch \bb4_ and \bm4_.  This is what the attached commit does.  Yes, it does mean that if someone uses b4_foo or m4_foo, she'll get a spurious warning.  If that's really a problem, I'll have to introduce a dedicated -W.  But that would lessen the interest of the warning, if most people don't enable it.

Hum...  Actually "escaping" the user's input to avoid the warning would be a better option.  I'll think about it.

Cheers!


commit c95d0dd5f54b98825e4cfe00a807db78a073e376
Author: Akim Demaille <[email protected]>
Date:   Sat Nov 6 19:24:04 2021 +0100

    warnings: be less picky about occurrences of m4_/b4_ in the output
    
    Reported by Marko Mäkelä.
    <https://lists.gnu.org/r/bug-bison/2021-10/msg00026.html>
    
    * src/scan-skel.l: It is ok to have foob4_ or foom4_.
    * tests/skeletons.at (Suspicious sequences): New.

diff --git a/THANKS b/THANKS
index be743a23..14e678c0 100644
--- a/THANKS
+++ b/THANKS
@@ -117,6 +117,7 @@ Marc Autret               [email protected]
 Marc Mendiola             [email protected]
 Marc Schönefeld           [email protected]
 Mark Boyall               [email protected]
+Marko Mäkelä              [email protected]
 Martin Blais              [email protected]
 Martin Jacobs             [email protected]
 Martin Mokrejs            [email protected]
diff --git a/src/scan-skel.l b/src/scan-skel.l
index 60232a2e..5a93ce03 100644
--- a/src/scan-skel.l
+++ b/src/scan-skel.l
@@ -54,7 +54,9 @@ static void output_mapped_file (char const *name);
 %}
 
 /* Identifiers of our M4 macros.  */
-macro    [bm]4_[a-zA-Z_0-9]*
+macro      [bm]4_[a-zA-Z_0-9]*
+/* Safe sequence of word-constituent characters.  */
+identifier [A-Za-z_0-9]+
 
 %x SC_AT_DIRECTIVE_ARGS
 %x SC_AT_DIRECTIVE_SKIP_WS
@@ -90,7 +92,7 @@ macro    [bm]4_[a-zA-Z_0-9]*
   /* This pattern must not match more than the previous @ patterns. */
 @[^@{}''(\n]*   fail_for_invalid_at (yytext);
 \n              out_lineno++; ECHO;
-[^bm@\n]+       ECHO;
+[^a-z@\n]+      ECHO;
 
   /* If there are still identifiers that look like macros, such as
      b4_synbol, this probably an error, say a typo in M4, or
@@ -104,6 +106,7 @@ macro    [bm]4_[a-zA-Z_0-9]*
                             "suspicious sequence in the output: %s", yytext);
                   ECHO;
                 }
+{identifier}    ECHO;
 .               ECHO;
 
 <INITIAL><<EOF>> {
diff --git a/tests/skeletons.at b/tests/skeletons.at
index 0ed11027..30f25a61 100644
--- a/tests/skeletons.at
+++ b/tests/skeletons.at
@@ -324,3 +324,44 @@ AT_SETUP([[Fatal errors but M4 continues producing output]])
 ]])
 
 AT_CLEANUP
+
+
+## ---------------------- ##
+## Suspicious sequences.  ##
+## ---------------------- ##
+
+AT_SETUP([[Suspicious sequences]])
+
+AT_DATA([[skel.c]],
+[[m4@&t@_include(b4_skeletonsdir/[c.m4])
+m4@&t@_divert_push(0)d@&t@nl
+@output(b4_parser_file_name@)d@&t@nl
+]b4_user_pre_prologue[
+]b4_user_post_prologue[
+b4_unevaluated
+m4@&t@_unevaluated
+]b4_epilogue[
+
+m4@&t@_divert_pop(0)
+]])
+
+AT_DATA([[input1.y]],
+[[%skeleton "./skel.c"
+%{
+  myb4_unevaluated
+  mym4_unevaluated
+%}
+%%
+start: ;
+%%
+myb4_unevaluated
+mym4_unevaluated
+]])
+
+AT_BISON_CHECK([[input1.y]], [], [],
+[[input1.tab.c:10: warning: suspicious sequence in the output: b4_unevaluated [-Wother]
+input1.tab.c:11: warning: suspicious sequence in the output: m4@&t@_unevaluated [-Wother]
+]])
+
+
+AT_CLEANUP