This is the mail archive of the
mailing list for the GCC project.
Re: [tree-ssa] Fix for failure to build glibc
- From: "Joseph S. Myers" <jsm28 at cam dot ac dot uk>
- To: law at redhat dot com
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Thu, 8 May 2003 23:25:01 +0100 (BST)
- Subject: Re: [tree-ssa] Fix for failure to build glibc
- References: <200305082146.h48LkJ3S013919@speedy.slc.redhat.com>
On Thu, 8 May 2003 email@example.com wrote:
> I suspect if tree-ssa can't compile current releases of the kernel,
> glibc and other key packages, then it will effectively be DOA. Simply
> saying that those packages are ill-formed isn't going to cut it IMHO.
In general new versions of GCC are never recommended for the Linux kernel,
and often don't work to compile it; and glibc releases may well not build
with compilers more recent than the glibc release (and, on the other side,
glibc changes may well stop older GCC releases from building, often in
libstdc++). If glibc is fixed now then a new release would probably come
long before any GCC release that makes jumping out of statement
expressions an error anyway.
There are extensions it makes sense to keep for the sake of glibc; various
constraint violations with flexible array members are only pedwarns if
pedantic, rather than unconditional hard errors, because the bad usages
are used in glibc, and __builtin_classify_type is retained (after
proposals to remove it as no longer used by GCC) because of use by glibc.
But jumping in and out of compound literals is just *so* badly defined
that I think requiring glibc to be fixed to build with newer (>= 3.4? >=
3.5?) GCC is the way to go.
Joseph S. Myers