Pre-release review of glibc 2.17: 0 P1 bugs, 14 P2 bugs, 0 P3 bugs.

[David Miller]

From: Mike Frysinger <vapier@gentoo.org>
Date: Thu, 29 Nov 2012 13:17:46 -0500

> On Thursday 29 November 2012 11:50:06 David Miller wrote:
>> From: "Carlos O'Donell" <carlos@systemhalted.org>
>> > "fclose violates POSIX 2008 on seekable input streams"
>> > http://sourceware.org/bugzilla/show_bug.cgi?id=12724
>> > - No patch exists, and too risky.
>> 
>> Agreed.  It's troubling to see that a major system component like Ruby
>> seems to be running into problems because of this.
> 
> various distros (at least Arch, Gentoo, and OpenSUSE) have been carrying this 
> revert patch since glibc-2.14:
> http://sources.gentoo.org/gentoo/src/patchsets/glibc/2.16.0/0068_all_glibc-2.16-
> glibc-revert-fseek-on-fclose.patch
> 
> *looks* fairly low risk

Thanks for the info.

Honestly, we are kidding ourselves if we don't do something about this when
upstream users of glibc are largely reverting the change.

The only entities testing and using the code as it is in GIT is us.

I think the easiest thing to do is revert the change.  That change
fixes a contrived test case, yet breaks known applications, and Ulrich
didn't seem interested at all in fixing the real regression he
introduced.