Re: ffi dependent packaqes

On Sat, Jun 22, 2002 at 11:14:19AM -0400, John DeSoi wrote:
> Hi Christophe,
> >I'm no expert, but I have worries about uffi's long-term viability in
> >its current state. I think my position on compatibility layers might be
> >causing me to be overly harsh, but I don't think that a solution to lack
> >of standardization is to take the current intersection of the
> >feature-sets and make it work on 32-bit platforms...
> 
> Could you provide a little more background on why you feel this way 
> about UFFI? I realize it is not a perfect solution, but I don't see 
> any other efforts aimed at unifying Lisp FFI. A consistent way to 
> access external libraries is a big detriment to Lisp. Are there any 
> better alternatives?

Firstly, let me say that I don't know of any better alternatives.
Secondly, let me say that the last thing I want to do is to stop people
from writing and exploring new code.

There are one or two "surface" things that stop me from liking uffi; let
me start with the most trivial first: some naming (yes, sorry :-). This
one's probably not an issue for some people, but I have a real problem
with DEF-TYPE and DEF-CONSTANT and DEF-STRUCT, because they are
pronounced the same as the CL macros DEFTYPE, DEFCONSTANT and DEFSTRUCT.

Told you that was trivial, but I generally think "audially" (though not
"audibly" :) so it's an annoyance.

Here's a somewhat more fundamental problem:

  :long - Signed 32-bits.
  :unsigned-long - Unsigned 32-bits.

There exist several platforms *today* where this isn't true, or, rather,
isn't useful.

I should say, though, that since at present I'm not involved in writing
any code that deals with foreign stuff I should be left on the side.

My other main concern reflects the fact that I come a little bit from a
"cathedral" mentality: there doesn't seem to be any justification that I
can find for the UFFI way of doing things other than "it works". While
that's laudable in itself, it shouldn't mean that UFFI becomes a defacto
standard (except for alphas, and openmcl, and clisp) just because of
that, in much the same way that I don't think the Allegro URI library
should or (shudder) clocc-port should.

I'll go back to trying to make sbcl's arrays less lamentable.

Cheers,

Christophe
-- 
Jesus College, Cambridge, CB5 8BL                           +44 1223 510 299
http://www-jcsu.jesus.cam.ac.uk/~csr21/                  (defun pling-dollar 
(str schar arg) (first (last +))) (make-dispatch-macro-character #\! t)
(set-dispatch-macro-character #\! #\$ #'pling-dollar)


-------------------------------------------------------
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/