Hello,
On Thu, Jan 31, 2013 at 10:30 AM, Thorsten Glaser <t...@mirbsd.de> wrote:
> there are testsuite failures (despite the nocheck build option),
> but the outputs provided look identical (WTF?).
That's the magic of floating-points operations ;-)... There has to
be a certain tolerance for comparing results, although in principle
the exact same operations are carried out (but compiler optimization
may make a mess out of that).
> One should add that m68k, like i8087, has 80 bits of internal
> precision, and this is not disablable on all current targets.
That's probably the reason.
> Full build log attached.
Thanks,
Vincent
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
