An lsbsi update
"Wichmann, Mats D" <[email protected]> Wed, 5 Jan 2005 16:48:14 -0800
| Newsgroups | gmane.linux.lsb.implementation |
|---|---|
| Message-ID | <A06801158AE07847B27A52C1A074BC1D06E92FB5@fmsmsx404.amr.corp.intel.com> |
Following on from earlier comments before the holidays, successful builds of the gcc-3.4-enabled lsbsi have now=20 taken place on four of the five architectures where it=20 has been attempted: ia32, ia64, x86_64, s390x. The one place where it's definitively failed is ppc64, which is being looked at by experts. I don't have access to a=20 ppc32 machine with space to try the build, and I haven't attempted it on s390. (the previously reported failed ia32 build "just worked" when I restarted it, it looks like some kind of system problem, not unknown on that system). The objective of this fiddling, as noted before, is to have a distro-neutral place to play with the C++ ABI=20 that is planned for LSB 3. On the four above noted=20 successful platforms, I've extracted the c++ symbols and compared them with the ones checked in to the gcc=20 tree (3.4 branch, under libstdc++-v3/config/abi), and=20 they only differ in a very small number of added symbols. Appears the checked in baseline is for gcc 3.4.0, and maybe a half-dozen things were added in the subsequent releases, which completely accounts for the variances. This is good news since we had to do extra work to verify things in the 3.3 timeframe, only some of the=20 architectures were updated for that release.=20