[Ecls-list] ECL 10.2.1 RC

Matthew Mondor mm_lists at pulsar-zone.net
Sat Feb 13 14:06:54 UTC 2010


On Fri, 12 Feb 2010 22:48:00 +0100
Juan Jose Garcia-Ripoll <juanjose.garciaripoll at googlemail.com> wrote:

> The current git tree is considered release candidate. If nobody speaks
> against it I will produce a formal release during the weekend.

I see the same issues as before with UFFI-compatible get-slot-value, I
suspect that it could possibly be an alignment related issue, but have
not confirmed this.  But I also realized that support for other C99
types, and some useful grovelling features were missing from UFFI.

However, after having written a few successful test macros to create C
structure getf/setf accessors, structure printer method, and to easily
map C constants to CL, I reverted to using the native c-inline FFI and
this route seems the sanest so far with ECL in my opinion.

This won't likely be ready for this release, but once my helper macros
are finalized, would it make sense to perhaps eventually include them
in ffi.lsp to extend the native c-inline FFI API?  An example I see is
contrib/sockets and contrib/serve-event which both define the same C->CL
constants helper macros, but could share the same code...  If that's
wanted, I could eventually post the proposed extensions here for review.

Of course, CFFI also has a good software base, supports grovelling and
can work in interpreted mode.  But for ECL-specific contributions the
native FFI will probably still remain the best choice, unless a native
CFFI support eventually exists (but it has various dependencies).

Thanks,
-- 
Matt




More information about the ecl-devel mailing list