This is the mail archive of the
mailing list for the glibc project.
Re: [PATCH] remove nested functions from regcomp.c
- From: "Carlos O'Donell" <carlos at redhat dot com>
- To: Will Newton <will dot newton at linaro dot org>, Konstantin Serebryany <konstantin dot s dot serebryany at gmail dot com>
- Cc: Roland McGrath <roland at hack dot frob dot com>, GNU C Library <libc-alpha at sourceware dot org>
- Date: Tue, 30 Sep 2014 10:44:14 -0400
- Subject: Re: [PATCH] remove nested functions from regcomp.c
- Authentication-results: sourceware.org; auth=none
- References: <CAGQ9bdyadYuhF4zzTve8Qn4YtD0shRGq3+Tz0yXSg1cQQDB+qw at mail dot gmail dot com> <CANu=Dmg33DOcA8bWhhWoYPeo9dAG8EYUayv_e+qLHub8LP+UcA at mail dot gmail dot com>
On 09/30/2014 04:13 AM, Will Newton wrote:
> On 29 September 2014 21:54, Konstantin Serebryany
> <firstname.lastname@example.org> wrote:
>> Please review the patch that removes nested functions from posix/regcomp.c.
>> The patch does not noticeably affect the generated code (same
>> instructions, a few differences in used registers, offsets, etc)
>> because all the affected functions have __attribute ((always_inline)).
>> No regressions in 'make check' on x86_64-linux-gnu (Ubuntu 14.04)
>> This code has some #ifdef branches for "not _LIBC",
>> please let me know if any separate testing is required for that case.
> This file is shared with gnulib so ideally the changes should be OKed
> there too. Unfortunately this particular file seems to have diverged a
> bit from gnulib so that might not be that straightforward.
The first step is to synchronize this file with gnulib.
This should be looked at first before considering any further changes,
however, Konstantin need not do this work himself, and we can help.
Lastly, we want *certainty* in our statements about compiler warnings
being wrong. A thorough reasoning needs to be brought forward to say
why the compiler warning is wrong. A hand-waving argument doesn't cut it.