On Tue, 01 Aug 2000, Jes Sorensen wrote:
> Just because it seems to compile the specific configuration you are
> using does NOT mean it will compile everything else correctly. Is that
> really so hard to understand?
No, that's evident. Still, if you don't try it because someone (you)
considers it unsupported, you won't fix it. Neither the kernel nor gcc.
I have compiled every so many Gigs of source code with gcc 2.95.2, and
the only time something broke was when AT&T's graphviz was nuked by
LC_COLLATE=de_DE.ISO-8859-1 that had a different notion of how to
sort spaces than LC_COLLATE=POSIX had. Ulrich Drepper claimed it was not
a libc bug, I'm not sure, I haven't read those ISO-1xxxx documents. AT&T
was notified, they were thankful for the hint and fix, everything is
peaceful.
However, I had some difficulties with older binutils alphas (what SuSE
6.4 delivers, 2.9.5.0.24 or whatever) bombing away on some C++ code,
which was fine with 2.9.5.0.42 or later. I also had gcc 2.95.2 bomb with
Internal Error, a bug in the "ice on illegal code" category, which is
fixed in the current gcc snapshot.
Still, gcc 2.95.2 has not yet trapped me although I use -O6
-march=pentium -mcpu=k6 on everything but the kernel, which sticks to
whatever the Makefile considers appropriate.
go use 2.95.2 and see where it bombs and fix the offender. Heck, gcc
2.95.2 cannot be all that buggy, why would FreeBSD 4.x work if it was?
-- Matthias Andree- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Mon Aug 07 2000 - 21:00:06 EST