Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!ucsd!helios.ee.lbl.gov!hellgate.utah.edu!hogum.utah.edu!clark
From: clark%hogum.utah.edu@cs.utah.edu (Charles Clark)
Newsgroups: comp.lang.lisp
Subject: akcl for ps/2 running AIX v1.2
Summary: raw_kcl runs... saved_kcl doesn't
Keywords: kcl akcl ps2 aix help
Message-ID: <1990Jul16.144255.13822@hellgate.utah.edu>
Date: 16 Jul 90 20:42:55 GMT
Sender: clark@cs.utah.edu (Charles F. Clark)
Followup-To: clark@cs.utah.edu
Distribution: usa
Organization: University of Utah CS Dept
Lines: 33
X-Local-Date: 16 Jul 90 13:42:55 PDT


An attempt to compile akcl v1.478 on an IBM ps/2 Model 70 running AIX v1.2 
has not been successful.  The raw_kcl generated seems to work correctly...
However, when the saved_kcl is reloaded for the second pass a segmentation
violation occurs.

The inclusion of the file 'rel_ps2aix.c' in the release suggests that this
version of akcl has been built on the ps/2.  Yet, no specific information,
in makefiles or otherwise, indicate what actions need to be taken.  

Any information would be most appreciated...

There are a couple changes I've made to get it to complete the first pass.

They include:

     (i)  converting 'earith.c' to assembly language  (the IBM C compiler
doesn't support the asm() statement...    well it does, but get this, the
argument required is not the usual string, but rather some number indica-
ting which operation is to be executed.  No documentation providing a map
from assembly to these numbers is provided by IBM with the compiler).

    (ii)  including the line '#define GETCWD' in the file 'att.h'.

   (iii)  commenting out all references to SFASL in the *.h files.  Seems 
there is a problem with the definition of ISCOFF and other magical stuff.

What remains is a problem with memory management when the saved_kcl is
loaded.

Thanks...

	--Charles (clark@cs.utah.edu)