[PATCH] [BZ #10631] Clarify malloc error diagnostics to avoid confusion

[Dmitry V. Levin]

On Tue, Sep 18, 2012 at 01:40:48PM +0530, Siddhesh Poyarekar wrote:
> On Tue, 18 Sep 2012 11:43:55 +0400, Dmitry wrote:
> > > 
> > > * ChangeLog should not be posted as part of the patch
> > 
> > It would mean manual handling of git-format-patch output, and the
> > result would not be git-am'able as is.  Well, if there is a strong
> > preference to have posted patches different from applied patches,
> > then OK.
> 
> The result may not necessarily be git-am'able in any case unless
> everyone has the changelog merge driver installed.

According to http://sourceware.org/glibc/wiki/GlibcGit,
git-merge-changelog is explicitly mentioned as a tool
"that can be used seamlessly to merge ChangeLog entries sensibly".

> Anyway, there is
> indeed a strong preference towards putting the ChangeLog entry
> separately and not as part of the patch, which is why we have it
> explicitly mentioned in the contribution checklist.

OK

> > > * Your submission needs a brief description of what you're trying
> > > to fix
> > 
> > A reference to [BZ #10631] is the brief description.
> 
> It is just more considerate towards reviewers to post a summary of the
> discussion here.

Yes, in a less trivial case a brief description would be appropriate.
In this case, the summary should be quite self-explanatory.

> > > >        __libc_message (action & 2,
> > > > -                     "*** glibc detected *** %s: %s: 0x%s ***\n",
> > > > +                     "*** glibc has detected an error in %s: %s:
> > > > 0x%s ***\n", __libc_argv[0] ?: "<unknown>", str, cp);
> > > >      }
> > > >    else if (action & 2)
> > > 
> > > The change is trivial but it may break scripts and diagnostic
> > > utilities that may be using the earlier format.
> > 
> > The message is for people rather than for automated processing by
> > scripts. Anyway, it is not part of libc ABI so the wording may
> > change.  More than that, some distributions already have this change
> > for several years, so any tool that rely on strict wording is already
> > unportable.
> 
> I don't have any strong preference for or against this change but I'll
> wait for an ack from another maintainer before I commit this for you.

Thank you, I can push the commit myself.


-- 
ldv
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/libc-alpha/attachments/20120918/44ac16b9/attachment.sig>