Re: [Fwd: libc6-dev: totally incoherent pthread related includes files for dynamic linking]
Eric Valette wrote:
The only release-relevant thing I can see here is that it's not a
release-critical bug. Thank you for bringing it to my attention so that
I could downgrade it. :-)
(The grave severity implies the package is unusable, which libc6-dev
isn't; it doesn't seem to fit any of the criteria for serious bugs in
the release policy, either.)
Well using include files belonging to another package/version is the
best way to create RC bugs that you will spend weeks to find. I'm _sure_
somebody did _indeed_ run some pthread related test to check the desired
behavior :-)
I decided to run the Open posix test suite 1.4.3 to check the level of
POSIX compliance of debian sarge and compared it with my own correctly
generated NPTL static libc from cvs on a chrooted sid partition. Here
are the final results :
--------------------------------------------------
| Original Libc6-dev | Up to date cvs (static) |
--------------------------------------------------
| PASS : 1239 | PASS : 1384 |
| FAILED : 28 | FAILED : 18 |
| UNTESTED : 46 | UNTESTED : 5 |
| UNRESOLVED : 16 | UNRESOLVED : 9 |
--------------------------------------------------
My analysis is :
- More test can be compiled because of using the correct includes
(especially the mq_* test) => more tests run (~150 more tests),
- Less failure due to bugfixes integrated in current glibc-cvs,
NB : probably the figures using the linuxtread static version of actual
libc-dev would be even far worse. The Original Libc6-dev tests by
default use libc-dynamic-nptl 2.3.2 as supplied by libc6 + libc6-dev.
=> when will debian upgrade its flaky libc :-)
--
__
/ ` Eric Valette
/-- __ o _. 6 rue Paul Le Flem
(___, / (_(_(__ 35740 Pace
Tel: +33 (0)2 99 85 26 76 Fax: +33 (0)2 99 85 26 76
E-mail: eric.valette@free.fr
Reply to: