> > > I reconfigured gmp so that it falls back to the pure C
> implementation,
> > > and the problem disappeared. Is the gmp stuff in ghc
> maintained at
> > > Glasgow, or are you syncing your code with the official
> distribution?
> >
> > It's an almost exact copy of the official distribution,
> with a few minor
> > tweaks to the configuration stuff, as I recall. Is the bug
> in the GMP
> > configuration, or in our build system?
>
> With NetBSD 1.3.2 and gmp-2.0.2, `./configure && gmake && gmake check'
> works fine. Since this is not the case in the ghc-4.08.2 source dist,
> I assume that it must be a ghc problem. Note that the targets differ
> slightly: it's `i386-unknown-netbsd1.3.2' in the originial and
> `i386-unknown-netbsd' in the rts. As I mentioned earlier going to
> ghc/rts/gmp and reconfiguring with `--target=unknown' does the trick.
It appears that our Makefile is passing the wrong target into the GMP
configure. We'll fix this, but in the meantime, you could either:
- remove the --target=$(HOSTPLATFORM) from ghc/rts/Makefile,
to let GMP figure out its own target, or
- install GMP yourself, so the GHC build system will detect that
you have it installed and not attempt to build its own version.
Thanks for the report.
Cheers,
SImon
_______________________________________________
Glasgow-haskell-bugs mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
