[Gcl-devel] GCL and Bill Schelter

[root]

Sorry if you've seen this before but Richard Fateman hinted that I might
not have copied everyone who cares...

Tim Daly
address@hidden
address@hidden

------- Start of forwarded message -------
To: Richard Fateman, Richard Stallman
From: Tim Daly
Subject: GCL and Bill Schelter


Richard[s],

I was the Axiom (nee Scratchpad) developer at IBM Research who worked
with Bill Schelter on AKCL. At the time AKCL was licensed first from
Kyoto thru Yuasa and Hagiya. Bill had an agreement that allowed him
to make changes atop the original (KCL). He built a mechanism that
merged the original with his own version of patch to construct AKCL.
IBM, in order to distribute Scratchpad, had a license with both the
KCL group and Schelter and had permission to distribute Scratchpad
with AKCL. I was part of those discussions.

At the time my efforts involved porting Scratchpad from MacLisp and
Lisp/VM to Common Lisp. Scratchpad eventually ran on any Common Lisp
including Symbolics, Golden Common Lisp, Lucid, Franz, AKCL, CMUCL
(Spice, actually, which later morphed into CMUCL), and IBUKI.

When Scratchpad was sold to NAG (Numerical Algorithms Group) as "Axiom"
the whole Axiom effort was replatformed from AKCL to CCL (Arthur Norman's
Codemist Common Lisp). The NAG version runs on CCL. Arthur Norman has
released a version of CCL under modified BSD and the sources are distributed
as part of the Axiom system. 

In the readme file Bill comments that the last version distributed
under the old license was akcl-1-624 which was, in fact, the last
version that Scratchpad used.


GCL may not in fact be a direct derivative of KCL. I know that Bill
had plans to rewrite the KCL pieces of the system and, given his
level of productive output, likely succeeded. Unfortunately we parted
ways once Axiom came out.

The basic build process involved compiling a file called "merge.c"
which was Bill's "patch" program. It took .V files and did context
sensitive replacements from KCL sources to AKCL sources. The .V files
no longer exist and there are no @s[ replacement instructions left.
The merge.c program still exists in the source tree but appears unused.
Thus I believe, but cannot prove, that he completely rewrote the KCL part.

Bill was extremely sensitive to licensing issues and we had numerous
discussions on the subject. I believe, knowing Bill, that he somehow
resolved the licensing issue with the KCL people. He was very careful
about the licensing issue. He was also deeply aware of the GPL issues
as he was a contributor to Emacs sources (look for his name in the
dbg handling under Emacs). It is unlikely that he included anything
without knowing he had permission.

As to the issue of distributing the improvements I believe, but can no
longer prove, that the IBM contract was written so that all changes
made by Bill were freely distributable. Scratchpad was a research
project and I know that, up until the issue of selling Scratchpad
started, I could freely distribute the Scratchpad sources if asked. I
know that various people have (or had) copies of the sources. The 
attitude at IBM Research (at least as I understood it) was very
close to the one Stallman expected which was "sources? sure. what
format would you like them in? tape or 5in floppy?". Bill had basically
the same attitude. What little money he made off AKCL was due to services,
not to selling source code, at least as far as I'm aware.

Actually, IBM did pay for the services. Bill was under contract
with IBM. At the time we had no plans to sell Scratchpad. We favored
Bill's AKCL because (a) Bill could help us port it to many platforms
(I wanted Scratchpad to run on everything, including DOS) and (b) Bill
was very receptive to helping us optimize Scratchpad as he was also
a user and contributor. Furthermore he was an excellent mathematician
so he could handle the complexity of Axiom's algebra.

The language you use to implement a system (such as Maxima) should not
be affected by the language implementation (GCL, Franz) license. At
some point you have to try to separate church and state. Isn't there
any way to be a programmer without becoming a lawyer also? Must I learn
about lache, estopple, waiver, and abandonment in order to program?
These days I feel like I should major in Law with a minor in CompSci.

Tim Daly
address@hidden
address@hidden

P.S. I'll take the $100,000 a year to maintain GCL :-)



_______________________________________________
Axiom-developer mailing list
address@hidden
http://mail.nongnu.org/mailman/listinfo/axiom-developer
------- End of forwarded message -------