This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] Simplify strncat.
- From: Joseph Myers <joseph at codesourcery dot com>
- To: David Miller <davem at davemloft dot net>
- Cc: <eggert at cs dot ucla dot edu>, <neleai at seznam dot cz>, <libc-alpha at sourceware dot org>
- Date: Wed, 17 Dec 2014 00:24:10 +0000
- Subject: Re: [PATCH] Simplify strncat.
- Authentication-results: sourceware.org; auth=none
- References: <20141216202438 dot GA5612 at domone> <54909B1E dot 9040201 at cs dot ucla dot edu> <20141216 dot 155247 dot 1722648331186689952 dot davem at davemloft dot net>
On Tue, 16 Dec 2014, David Miller wrote:
> The fact that we are talking about complete rewrites for what should
> be a one line warning fix shows that we enabled -Werror prematurely.
No, it simply shows people getting side-tracked and confusing the
questions of how to fix the build for an existing warning and how to clean
up the issue it shows up. The variety of different configurations in
which people build glibc fundamentally requires the process of fixing
warnings to be done in parallel by people seeing those warnings (even
restricted to x86_64 with GCC 4.9, some people are reporting different
warnings I don't see in such a configuration, so testing all combinations
of architectures and compiler versions would not cover the warning space).
The process of fixing warnings started while the -Werror policy was under
discussion (there was plenty of opportunity for people to fix warnings
they saw in advance of -Werror being enabled - if they didn't take that
opportunity, announcing "-Werror will be enabled at X time in the future"
would hardly have resulted in much more fixing).
My position remains as in
<https://sourceware.org/ml/libc-alpha/2014-11/msg00693.html>:
I think we should be perfectly willing to file a bug in Bugzilla and then
add a suppression referencing the bug, if there's something tricky about
determining if there's a real glibc bug there or a better way to address
the warning - that means the build stays working on more platforms (in the
presence of -Werror). I see no reason why we should assume a possible
problem in existing code shown up by a warning (from new GCC or on another
architecture) is more serious than other bugs. -Werror is first about
stopping new issues getting in accidentally, rather than forcing certain
existing issues (those that cause warnings) to be higher priority than
other existing issues.
(If the warning that it's hard to work out the correct fix for arises from
a recent change to glibc, we should be rather more willing to revert that
change until the fix has been worked out, although that's not necessarily
the right thing to do in all cases.)
--
Joseph S. Myers
joseph@codesourcery.com