[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: gcc on Irix 6.5
From: |
Bruno Haible |
Subject: |
Re: gcc on Irix 6.5 |
Date: |
Fri, 27 Feb 2009 16:31:32 +0100 |
User-agent: |
KMail/1.9.9 |
Eric Blake wrote:
> > but this is due to a gcc bug: If floorl(-0.7L) is executed before
> > floorl(-1.0L), and both calls are compiled by gcc - regardless whether
> > in the same function, in different functions, or even in different
> > compilation
> > units -, then the floorl(-1.0L) call returns -2L instead of -1L. If at
> > least one
> > of the calls is in a compilation unit compiled by cc, no problem.
>
> What a strange bug! Obviously, older gcc on Irix is not the best option;
Well, the newer gcc 3.4.3 that you are using is also having problems - that the
older
one did not have:
../../gltests/test-frexpl.c:144: assertion failed
../../gltests/test-ldexpl.c:130: assertion failed
> maybe autoconf should default to cc on this platform, rather than always
> favoring gcc.
I disagree. I've built many packages with that compiler on IRIX, without
problems.
The problem is more with the "long double" area - which is not much used by
other
program, hence less reliable - than with gcc. Any kind of rarely used features,
be
it global register variables, OpenMP support, or similar, has more bugs than
average
features.
> Should we update install.texi (and thus INSTALL) to mention
> that on Irix, if gcc doesn't work, to try cc?
Certainly not. It would make people doubt of gcc's quality. And gcc - especially
since version 4.0 - is of outstanding quality and reliability.
Bruno
- computing long double NaN on Irix 6.5, Eric Blake, 2009/02/26
- Re: computing long double NaN on Irix 6.5, Eric Blake, 2009/02/26
- Re: computing long double NaN on Irix 6.5, Bruno Haible, 2009/02/27
- spurious roundl output (was: computing long double NaN on Irix 6.5), Eric Blake, 2009/02/27
- Re: spurious roundl output (was: computing long double NaN on Irix 6.5), Bruno Haible, 2009/02/27
- Re: spurious roundl output (was: computing long double NaN on Irix 6.5), Eric Blake, 2009/02/27
- Re: spurious "(cached)" output from AC_CACHE_VAL, Bruno Haible, 2009/02/27
- Re: spurious "(cached)" output from AC_CACHE_VAL, Eric Blake, 2009/02/28
- Re: spurious "(cached)" output from AC_CACHE_VAL, Ralf Wildenhues, 2009/02/28
- Re: computing long double NaN on Irix 6.5, Eric Blake, 2009/02/27
- Re: gcc on Irix 6.5,
Bruno Haible <=
Re: computing long double NaN on Irix 6.5, Tom G. Christensen, 2009/02/27