[Bug bootstrap/35855] build locale not properly handled with awk scripts

[Bug bootstrap/35855] build locale not properly handled with awk scripts

[rwild at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35855

--- Comment #2 from Ralf Wildenhues <rwild at gcc dot gnu.org> 2010-09-26 06:07:56 UTC ---
Proposed patch at <http://gcc.gnu.org/ml/gcc-patches/2010-09/msg02027.html>.

You don't state how exactly the build fails, and what locale you needed to set
for the failure to become apparent.  That makes it harder to find out whether a
patch fixes it.  Please state so, and also please try out the patch.  Thanks.

[Bug bootstrap/35855] build locale not properly handled with awk scripts

[vapier at gentoo dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35855

--- Comment #3 from Mike Frysinger <vapier at gentoo dot org> 2010-09-26 20:16:07 UTC ---
the same locale that breaks every time: et_EE

example build failure:
http://bugs.gentoo.org/215828

[Bug bootstrap/35855] build locale not properly handled with awk scripts

[rwild at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35855

--- Comment #4 from Ralf Wildenhues <rwild at gcc dot gnu.org> 2010-10-11 17:58:56 UTC ---
Author: rwild
Date: Mon Oct 11 17:58:53 2010
New Revision: 165322

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=165322
Log:
Fix PR bootstrap/35855: awk character classes.

gcc/:
    PR bootstrap/35855
    * opt-functions.awk (BEGIN): New section.
    (lower, upper, digit, alnum): New variables.
    (static_var, opt_sanitized_name): Use alnum instead of character
    classes, for non-C locale.
    * optc-gen.awk: Likewise.
    * opth-gen.awk: Likewise.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/opt-functions.awk
    trunk/gcc/optc-gen.awk
    trunk/gcc/opth-gen.awk

[Bug bootstrap/35855] build locale not properly handled with awk scripts

[rwild at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35855

Ralf Wildenhues <rwild at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
      Known to work|                            |4.6.0
         Resolution|                            |FIXED

--- Comment #5 from Ralf Wildenhues <rwild at gcc dot gnu.org> 2010-10-11 18:58:27 UTC ---
Fixed in trunk.

[Bug other/35855] New: build locale not properly handled with awk scripts

[vapier at gentoo dot org]

the gcc build system has some awk scripts that use unsafe character ranges:
$ grep a-z gcc/*.awk
gcc/optc-gen.awk:       gsub( "[^A-Za-z0-9_]", "X", macros[i] )
gcc/optc-gen.awk:       gsub ("[^A-Za-z0-9]", "_", enum)
gcc/opt-functions.awk:  gsub ("[^A-Za-z0-9]", "_", name)
gcc/opth-gen.awk:       gsub( "[^A-Za-z0-9_]", "X", macros[i] )
gcc/opth-gen.awk:       gsub ("[^A-Za-z0-9]", "_", enum)

A-Z will not match the expected alphabet (the range as defined by the "C"
locale) in all locales.  while this has always been a problem, it went
unnoticed as the incorrect munging is consistent in nature.  with gcc-4.3, the
incorrect munging results in a conflict of symbols and triggers a build
failure.

i cant seem to find any precedent as to the correct fix.  i would personally
just change everything to [:alnum:], but the only use of such character classes
that i can find via a quick grep is in the lex files.  another solution would
be to execute the awk stuff via configure as it sets up a clean environment. 
or you can prepend "env LC_ALL=C" to the AWK variable setup via configure, but
this ignores the problems that configure solves automatically.


-- 
           Summary: build locale not properly handled with awk scripts
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: other
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: vapier at gentoo dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35855

[Bug bootstrap/35855] build locale not properly handled with awk scripts

[urmet dot saar at gmail dot com]

------- Comment #1 from urmet dot saar at gmail dot com  2009-03-10 17:24 -------
I can confirm this, it's been annoying me for some time.
When I changed every "A-Za-z0-9" to [:alnum:] the symbol conflicts went away
and diff confirmed that the generated files were identical to the ones
generated earlier with LC_ALL="C"


-- 

urmet dot saar at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |urmet dot saar at gmail dot
                   |                            |com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35855