Re: Defining a libffi.so.4 ABI

[Mladen Turk <mturk at redhat dot com>]

On 03/11/2010 03:28 PM, Rainer Orth wrote:
> "M.C.A. (Marco) Devillers"<marco.devillers@gmail.com>  writes:
>
> > Well, if you would be redefining a new  API, there are some things I
> > would like (out of the top of my head):
> > a) Treat the return type as part of the CIF. (Why not have all the
> > static information in it?)
> > b) Don't treat a void type different than any other type. (Don't
> > really remember, it led to a quirk in my compiler).
> >
> > I think if you do that, you'll find it becomes a lot more orthogonal.
>
> Let me repeat another time: I'm not interested in changing or redefining
> the libffi API in any way.  That's for the libffi maintainers to pursue.

Agree. Current API only needs FFI_EXPORT that would in a similar
way to FFI_HIDDEN determine the public API.
With that, FFI_HIDDEN is actually not needed any more cause anything
not FFI_EXPORT is by default hidden.


Regards
--
^TM