Re: Patch: support non-libffi enabled platforms

[Alexandre Oliva <aoliva at redhat dot com>]

On Feb 15, 2001, "Anthony Green" <green@redhat.com> wrote:

> Tom wrote: 
>> However, couldn't we somehow ask libffi whether it should be used?
>> That way we only need to do the configure hacking in a single place.
>> Ideally libffi could tell us (1) if it works on the platform and (2)
>> if the closure code is supported on the platform.

> What do you think the best way of doing this is?

Have libffi never fail to configure.  If it's to target an unsupported
platform, just create a libffi-config.h that says it's unsupported.
If it's on a supported target, have libffi-config.h tell whether
closure code is supported or not.  Then, have configure tests that
#include libffi-config.h and decide whether to not use it at all, to
use it in a limited form or to use all of it.

If we went ahead and created an empty library for libffi even on
unsupported platforms, we wouldn't even need configure tests; it'd be
enough to arrange for all of libffi users to #include libffi-config.h
and disable the usage of features according to the macros defined
there.  However, it might be better to have configure tests anyway; it
probably doesn't make sense to build a non-functional interpreter in
case libffi is totally unsupported.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me