On Wed, Apr 09, 2008 at 07:16:53PM +0000, Joseph S. Myers wrote: > On Wed, 9 Apr 2008, Stephen R Marenka wrote: > > > I'm not liking the feeling that our old libc and new gcc-4.3 > > (gcc-4.3_4.3.0-2) aren't getting along well. > > > > I have probably 20 failed packages, of which the following are > > representative examples. > > > > They all are compiled with -std=gnu99. It looks like that construct > > worked fine for gcc-4.2. > > > > Anyone care to follow this up? > > > > Makes me wonder if we can compile a modern glibc without TLS. I hope > > someone reading this is feeling called to be the glibc maintainer for > > m68k. Otherwise, I'm not sure we're going to get out of this. > > The glibc support for the C99 inline semantics was added in glibc 2.6 (via > a large 2007-03-16 patch of Jakub's and some followups), and 2.6 is also > the version where all the non-TLS support was removed. GCC 4.3 and later > use C99 inline semantics in -std=gnu99 mode. To use any earlier version > of glibc with GCC 4.3 you need at least to backport the large patch. I guess now would be a good time for TLS to become available. :) So we need to either keep a few buildds at gcc-4.2 to handle the -std=gnu99 packages or we need to modernize our glibc, perhaps even both. Meanwhile, should we start failing or dep-waiting all these packages? I'm inclined to depwait them on libc6-dev (>= 2.6), what do ya'll think? -- Stephen R. Marenka If life's not fun, you're not doing it right! <stephen@marenka.net>
Attachment:
signature.asc
Description: Digital signature