From: Bill Bouma
Subject: CMUCL foreign functions on SGI
Date: 
Message-ID: <3ri221$6b4@deneb.cs.purdue.edu>
The foreign function interface on SGIs is broken in CMUCL. (On Sun it
works fine.)  We are wondering if anyone is working to fix this, and 
when it will be done?  If not, we will have to buy more Franz.  Thanks
for any info.
-- 
Bill <·····@cs.purdue.edu>            
http://www.cs.purdue.edu/people/bouma  --  The worm gets the late bird.
From: Mike McDonald
Subject: Re: CMUCL foreign functions on SGI
Date: 
Message-ID: <3ri5or$8in@fido.asd.sgi.com>
In article <··········@deneb.cs.purdue.edu>, ·····@cs.purdue.edu (Bill
Bouma) writes:
|> 
|> The foreign function interface on SGIs is broken in CMUCL. (On Sun
|> it
|> works fine.)  We are wondering if anyone is working to fix this, and
|> |> when it will be done?  If not, we will have to buy more Franz. 
|> Thanks
|> for any info.
|> -- 
|> Bill <·····@cs.purdue.edu>            
|> http://www.cs.purdue.edu/people/bouma  --  The worm gets the late
|> bird.

  I'm trying but no promises at this point. Right now, I'm trying to get
CMUCL to
build from scratch on a SGI. I figure that's the hardest part. The FFI
looks
straight forward at this point in time. I'll let everyone here know if
and when I
get it finished (or when I get stuck).

  Mike McDonald
  ·······@engr.sgi.com
From: Liam Healy
Subject: Re: CMUCL foreign functions on SGI
Date: 
Message-ID: <3rkrlo$8e0@ra.nrl.navy.mil>
I've been looking into this.  It's easy to modify
code/foreign.lisp to use dynamic loading for foreign calls,
as is done with the Solaris version -- using dlopen and
dlsym.  Unfortunately, even once this is done, foreign
functions don't work, because there's some problem in the
compiler code generation (you always get a SIGBUS error: dlopen and
dlsym always return their own addresses instead of the result of calling them).  
I thought I had located the problem, but alas not.  
I'm still working on it; if I come up with anything, I'll post.

Liam Healy
··········@nrl.navy.mil


·······@engr.sgi.com (Mike McDonald) wrote:
>In article <··········@deneb.cs.purdue.edu>, ·····@cs.purdue.edu (Bill
>Bouma) writes:
>|> 
>|> The foreign function interface on SGIs is broken in CMUCL. (On Sun
>|> it
>|> works fine.)  We are wondering if anyone is working to fix this, and
>|> |> when it will be done?  If not, we will have to buy more Franz. 
>|> Thanks
>|> for any info.
>|> -- 
>|> Bill <·····@cs.purdue.edu>            
>|> http://www.cs.purdue.edu/people/bouma  --  The worm gets the late
>|> bird.
>
>  I'm trying but no promises at this point. Right now, I'm trying to get
>CMUCL to
>build from scratch on a SGI. I figure that's the hardest part. The FFI
>looks
>straight forward at this point in time. I'll let everyone here know if
>and when I
>get it finished (or when I get stuck).
>
>  Mike McDonald
>  ·······@engr.sgi.com
comp.lang.lisp / 1995 / jun