Path: utzoo!attcan!uunet!mcvax!ukc!etive!aiva!jeff
From: jeff@aiva.ed.ac.uk (Jeff Dalton)
Newsgroups: comp.lang.lisp
Subject: Re: foreign functions to KCL
Message-ID: <661@aiva.ed.ac.uk>
Date: 15 Nov 88 18:38:35 GMT
References: <5363@medusa.cs.purdue.edu> <6381@june.cs.washington.edu>
Reply-To: jeff@uk.ac.ed.aiva (Jeff Dalton)
Organization: Dept. of AI, Univ. of Edinburgh, UK
Lines: 32

In article <6381@june.cs.washington.edu> whaley@uw-june.UUCP (Ken Whaley)
writes:
>The good news is that it's easy to incorporate C functions into
>KCL.  The bad news is that unless you want to write C code that can't
>be used outside of the KCL environment, then you are constrained to use
>certain input parameter and function return types.

That's more or less true if you want to return something to be manipulated
in Lisp -- it has to be one of the objects Lisp knows about.  However, you
can malloc things, define (Lisp) procedures that call C procedures to
manipulate such objects, and so on.  It's just that there isn't a "C
structure" facility built-in.

>:NOTE:  The KCL documentation says that the following is specific
> to the BSD version of KCL.

I think only some it it (such as si:faslink) is just for BSD.  But since
I almost always use BSD I can't say for sure that it works.  Nonetheless,
since KCl compiles to C, some kind of C interface is surely possible.

>Oh, and I've only managed to get this working with non-pointer types.
>I haven't experimented with structs, either.  You can pass in, manipulate,
>and return arbitrary LISP objects (KCL itself is written in C....), but
>that sort of information is, as far as I know, undocumented, and you'd
>have to dig through the source code to figure it out.

Basically, Lisp objects are pointers to a union of structs.  So each
Lisp object corresponds to some C struct.  The structs, union, etc.
are defined in object.h and the subset used by compiled code is in
cmpinclude.h.

-- Jeff