GMP 5.0.5 build fails with tcc due to x86_64 asm (unknown register...)
paul at leyland.vispa.com
Wed Jun 20 18:40:15 CEST 2012
On Wed, 2012-06-20 at 11:41 +0200, Vincent Lefevre wrote:
> On 2012-06-20 11:13:19 +0200, Torbjorn Granlund wrote:
> > Vincent Lefevre <vincent at vinc17.net> writes:
> > Not to remove it, but to make it *optional*, i.e. to have a fallback
> > in C.
> > Excellent proposal. Implemented in GMP since the beginning...
> > Here it doesn't work with tcc. If you had a fallback in C, things
> > would be fine.
> > Please see the manual for how to exercise the fallback.
> But why isn't this detected by configure?
> > (Silently disabling assembly for incapable compilers seem like a very
> > poor idea.)
> Well, perhaps. But I think that a configure test + an error message
> (e.g. "Your compiler does not support the assembly code required by
> GMP. You can configure with --host=none to disable assembly code.")
> would be better than an error with "make".
FWIW, this is The Right Thing to do in my view.
It's friendly, helpful, needs implementing only once and gives a strong
hint that the person might do better to use a more powerful compiler if
the explicit advice is not followed for whatever reason.
More information about the gmp-discuss