Re: patch that fixes AC_COMPILE_IFELSE for Erlang tests

[Paolo Bonzini]

On 08/12/2009 07:28 AM, Ralf Wildenhues wrote:
> * Romain Lenglet wrote on Sat, Aug 08, 2009 at 04:24:04AM CEST:
> > --- a/lib/autoconf/general.m4
> > +++ b/lib/autoconf/general.m4
> > @@ -2557,7 +2557,7 @@ AC_DEFUN([AC_EGREP_HEADER],
> >   # Shell function body for _AC_COMPILE_IFELSE.
> >   m4_define([_AC_COMPILE_IFELSE_BODY],
> >   [  AS_LINENO_PUSH([$[]1])
> > -  rm -f conftest.$ac_objext
> > +  rm -f conftest.$ac_objext conftest.beam
>
> It seems a bit ugly that for languages other than Erlang, this change
> should be expanded as well.  Likewise for the other changes below.
> IIUC the Right Way would be to have a per-language $ac_objext, but that
> is much more than I'm willing to tackle right now (and would likely
> cause backward compatibility issues); but can we make these changes only
> if the current active language is Erlang?

I think we should not include this patch in 2.65, and possibly tackle 
per-language AC_COMPILE_IFELSE afterwards.  (Just like we should not 
include any m4sugar changes; there's nothing specific about Erlang).

My cross-compilation patch is bigger than what I'd like to have in 2.65, 
but may be an exception because it fixes a regression.

Paolo