[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gcl-devel] UFFI

From: Steven H. Rogers
Subject: Re: [Gcl-devel] UFFI
Date: Mon, 27 Jun 2005 21:05:59 -0600
User-agent: Mozilla Thunderbird 1.0 (X11/20041206)

Hi Camm:

Camm Maguire wrote:
Greetings, and thanks for your interest in this!

You're welcome.

"Steven H. Rogers" <address@hidden> writes:

From reviewing the archives, it seems that while Camm thinks porting UFFI to 
GCL would be a good thing to do and should
be straight forward, he's too wrapped up with ANSI compliance for 2.7 to 
address it in the near term.  Is anyone else
working on UFFI for GCL?

I do not know of anyone, and I will not have time until ansi is done.

GCL has a well-developed and powerful foreign function interface, but
not the standard uffi one (yet).  It is based aournd defcfun,
defentry, and clines (see docs, and threads in gcl-devel for examples
re: blas library interface).  What is needed is someone who has read
up on what uffi requires, and can map this functionality to that spec.

If no one is working on this and one were begin this project in the near 
future, would it be best to work with GCL 2.6.6
or 2.7.0t1?

If uffi requires ansi semantics, I'd use the latest testing version,
which is now Version_2_7_0t2.  Otherwise, I'd use 2.6.6, as nothing
we're doing in 2.7 is likely to affect what is needed for uffi, and it
represents a more stable starting point.

Are you volunteering (hope so)?  Can you indeicate what end-user (as
opposed to developer) applications this might enable on gcl?

I'm tempted to volunteer, but not quite ready to commit. I've begun looking at the UFFI docs and will compare them to GCL's foreign function facilities. I suspect that UFFI will be a functional subset of GCL's FFI, so mapping shouldn't be too difficult.

Applications which interest me:

from http://uffi.b9.com/links.html
- clsql
- cl-pdf
- cl-syslog

- cells-gtk http://common-lisp.net/project/cells-gtk/

There's a thread on c.l.l about the impedance mismatch between Unix and Common Lisp. I've been thinking about this, and UFFI seems like a way to reduce this mismatch.

Steven H. Rogers, Ph.D., address@hidden
Weblog: http://shrogers.com/weblog
"He who refuses to do arithmetic is doomed to talk nonsense."
-- John McCarthy

reply via email to

[Prev in Thread] Current Thread [Next in Thread]