libffi merge

Ralf Wildenhues
Sun Jun 7 06:44:00 GMT 2009

* Andrew Haley wrote on Sat, Jun 06, 2009 at 10:21:12AM CEST:
> NightStrike wrote:
> > 
> > Wouldn't it be ideal for the libffi in the gcc tree to be identical to
> > upstream, so that this mess doesn't occur again?
> Of course it would, but the configury is not different just for the sake
> of it, it's because the requirements are different.  I don't want there
> to be any differences at all, but we have to do what is possible.

I think we can fix that mostly. differences can use
m4-time conditionals (m4_ifdef([IN_GCC], [...], [...]) or so) with
a suitable define, and multilib differences in the files
can *hopefully* be factored under an automake conditional.  At least
that could push differences to a few bits.  Would standalone libffi
be willing to accept into its tree configury parts that are only
active in the GCC tree?

I'd look into this more but I really would like to be able to update
Autoconf and Automake used in the GCC tree first, that will likely make
things a bit easier.


More information about the Java-patches mailing list