Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!cmcl2!rutgers!rochester!ritcv!eer
From: eer@ritcv.UUCP (Ed Reed)
Newsgroups: comp.lang.lisp,comp.unix.wizards
Subject: Re: KCL on 3B machines?
Message-ID: <717@ritcv.UUCP>
Date: Sat, 17-Oct-87 08:28:11 EDT
Article-I.D.: ritcv.717
Posted: Sat Oct 17 08:28:11 1987
Date-Received: Sun, 18-Oct-87 11:08:28 EDT
References: <3@unf5.UUCP>
Reply-To: eer@ritcv.UUCP (Ed Reed)
Organization: Rochester Institute of Technology, Rochester, NY
Lines: 57
Keywords: KCL 3Bs success
Xref: mnetor comp.lang.lisp:500 comp.unix.wizards:4967

In article <3@unf5.UUCP> layne@unf5.UUCP (Layne Wallace) writes:
>Has anyone brought Kyoto Common Lisp up on a 3B machine? I keep getting an
>error along the lines of:
>
>     bad bigsize: 012
>
Let's be more specific - when you say 3B machine, do you mean...

	Unix PC (7300)
	3B1
			(both those are 68010 machines from 
			 Convergent Technologies, with CT's Sys V)
	3B2
			(family of WE32000 chip computers from ATT)

There's a difference.  The distribution tape we got was for the
3B2, and except for some .h file errors in the ATT supplied
include libraries, was almost ok.  Took a day or so to
track down all the little discrepancies.  One noteable prblem was
that the linker has an upper limit to the string table size in the
bound unit, which was exceeded under some linking circumstances - I've
forgotten now just how we got around that...maybe someone else
can fill you in.

I'm presently finishing porting Kyoto to the 3B1 family machines
and have the interpretor running just fine - big issue there was that
the Convergent linker padds segment to nice even starting offset
addresses in the a.out file, and correctly documents that fact in
the section headers, but the unixsave.c routine in the 3B2 sources
didnot allow for that - had to rewrite unixsave to work.

The only problem I'm having now is getting compiled code to fasl
correctly - looks like I've got some more pointer/offset problems
that I havent tracked yet - I can load init files okay (lisp code
compiled and distributed as c source that I only have to load the
lisp data vector from) but when I write my own lisp function "foo"
(well, "fib", actually) and compile it with my own running compiler
and then load the resulting object file - it blows up trying to 
execute the init code generated by the compiler for my function.

The other advise for you if you're working with 3B1 machines - 
20MB is too little.  Need at least the 40 MB drive to have enough
room to develop/debug Kyoto.

Good luck.  I'd also be interested in hearing from anyone else
working on the 3B1 port.

Ed Reed

-- 
-------------------------------------------------
Ed Reed - Rochester Institute of Technology
phone:    (716) 334-3006
Delphi:   EERTEST
GEnie:    SQA.INC
Usenet:   ...rochester!ritcv!eer
-------------------------------------------------