This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: Test failures on HP-UX 10.20
- From: Brian Gough <bjg at network-theory dot co dot uk>
- To: gsl-discuss at sources dot redhat dot com
- Date: Wed, 21 Nov 2001 10:55:35 +0000 (GMT)
- Subject: Re: Test failures on HP-UX 10.20
- References: <20011120113900.A7206@oolong.il.thewrittenword.com>
gsl-discuss@lists.thewrittenword.com writes:
> Against 1.0 compiled with:
> CC=cc (HP-UX commercial C compiler)
> CFLAGS="+O2 -Ae -z +ESlit +Onofltacc +DAportable +Oentrysched \
> +Odataprefetch"
> ...
> FAIL: gsl_isinf(inf) (0 observed vs 1 expected)
> FAIL: gsl_isinf(-inf) (0 observed vs -1 expected)
> FAIL: gsl_isnan(nan) (0 observed vs 1 expected)
> FAIL: gsl_finite(inf) (1 observed vs 0 expected)
> FAIL: gsl_finite(nan) (1 observed vs 0 expected)
Thanks for the bug report. The compiler is probably optimising away
tests for nan like (x != x) in sys/infnan.c. I don't have access to
that platform to investigate it or try out any workarounds. On IRIX
the option -OPT:IEEE_NaN_inf=ON was needed to ensure correct infinity
arithmetic for the same reasons.
Brian