[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
fugue # 2
- To: Kim.franz-friends@Berkeley
- Subject: fugue # 2
- From: Kim.fateman@Berkeley
- Date: Sat, 30 Oct 82 18:51:58 GMT
- Original-date: 30 Oct 1982 11:51:58-PDT
FUGUE Notes
An occasional publication of the
Franz Lisp User Group under Unix and Eunice (FUGUE)
Number 2 (October, 1982)
edited by Richard J. Fateman
University of California
Berkeley CA 94720
USA
fateman@berkeley
1. Welcome!
It seems about time to publish the second of these
newsletters, since we have accumulated a number of new
items. We would also like to relay to others such informa-
tion as has been forwarded to us. The reports of projects at
Berkeley (and elsewhere) may strike sympathetic chords with
other research.
2. New programs
2.1. OPS-5
OPS-5 is a "production system" written by Charles Forgy
of CMU. It appears to work just fine in Franz, and is in
wide use. Interested persons may obtain copies of documen-
tation and the program from Charles.Forgy@CMU-10A. ( Charles
Forgy, Computer Science Department, Carnegie-Mellon Univer-
sity, Pittsburgh, PA 15213)
It is their policy to send it to anyone who wants it free of
charge.
2.2. GLISP
GLISP is a system which provides interesting linguistic
features for generic operations and data abstraction. Writ-
ten by Gordon Novak at Stanford University, it was origi-
nally developed for Interlisp, but has been ported to other
lisps, including Franz.
2.3. Flavors
There are now two distinct implementations, apparently
with identical functionally, of "flavors" as appearing in
the MIT Lisp Machine software. One is described in TR-1174,
____________________
9 UNIX, Eunice, Franz Lisp, may be trademarks of Bell Labs,
SRI Int'l, and Univ. of Calif.
9
"Franz Flavors" by Richard J. Wood (Dept of C.S., Univ. of
Maryland, College Pk, MD 20742). The other was written by
Juan R. Loaiza of MIT, Laboratory for Computer Science. We
have a copy of the latter on-line here, and expect to
receive a copy of the Maryland one, shortly. Eric Cooper
here at Berkeley is in charge of the flavors situation.
There is an implementation of closures, mostly compati-
ble with the Lisp Machine specification, announced by John
Foderaro for Opus 38.33. The incompatibility is a result of
what we perceive to be a high performance penalty for eso-
terica.
2.4. Database Interfaces
Jim Larus at UCB has cooked up interfaces to both the
INGRES relational database system, and the simpler TROLL
database system. These will be described in his forthcoming
MS report, along with the next item.
2.5. Cursor-control and Menus
Larus has provided an implementation of screen manage-
ment which uses the UNIX "curses" package for primitive win-
dow management. A menu-based interface has also been
developed as part of this.
2.6. Vaxima and Algebraic Manipulation
A new version of vaxima, the VAX version of the MACSYMA
algebraic manipulation system, was released in July by UCB,
incorporating some bug fixes, improved programs, and a large
number of user-contributed subroutine libraries. This was
made available to test-site licensees. Unfortunately, MIT
has suspended new test-site licensing since about April,
1982. We hope that MIT will be liberalizing its distribu-
tion policy to non-commercial sites.
See the note below about MACSYMA being sold.
As a counterpoint to this, UC Berkeley has received a
substantial grant from the System Development Foundation for
work on Mathematical Representation and Manipulation, which
should result in some more advanced systems for application
of computers to symbolic mathematics. Recruiting for
researchers, staff, and students is underway now, and
interested persons should contact Richard Fateman.
2.7. VLSI Design Rule Checker
Lyra, written in Lisp by Michael Arnold, is a retarget-
able, hierarchical, design rule checker for VLSI circuits.
Lyra features a rule compiler (also written in Lisp of
course!) which translates symbolic design rule descriptions
to lisp code for checking the rules. Lyra was used for the
RISC project. It is currently being used extensively at
Berkeley, and will be included in the Fall-82 distribution
of of the Berkeley CAD tools. For more information contact
Michael Arnold or John Ousterhout at Berkeley.
2.8. Generic Arithmetic
As a proposed extension to Franz arithmetic, Richard
Fateman, Keith Sklower and Scott Morrison, have written a
simple-minded generic arithmetic package which includes
modules which can be loaded to support exact rational arith-
metic, software-simulated IEEE extended arithmetic, arbi-
trary precision floating point, complex, interval, and mul-
tivariate polynomial. Combinations of some of these are sup-
ported, although the package is as yet incomplete in some
areas. The IEEE arithmetic simulation is written in C.
These packages are probably not in good enough shape for
casual use by others.
3. New features
Various performance enhancements and bug fixes have
been incorporated in versions of Franz (now on Opus 38.33
and the compiler, Liszt 8.14) These are mentioned in brief
here; more details accompany updates of the system and
manual included in the forthcoming Berkeley 4.2BSD UNIX dis-
tribution.
3.1. Franz
We added a switch to cause the evaluator to save macro
expansions so they need only be expanded once.
We added vector and vector-immediate data types.
We rewrote showstack and backtrace so they are easier
to use.
We made the lisp to foreign function interface more
secure. The system now allows foreign function to call lisp
functions.
We added closures and support flavors, features from
the Lisp Machine.
3.2. Liszt
Liszt will check the number of arguments to system
functions and user defined functions.
9
9
Liszt supports local declarations.
Liszt will automatically compile lambda expressions
headed by the function `function'.
Liszt supports compiler-only macros and will autoload
macros if necessary.
4. MC68000
Keith Sklower and Kevin Layer have been working on the
MC68000 version of Franz under the UNIX operating system
(using a DUAL System 83). While the current configuration is
a swapping system, the Lisp should be able to use the full
address space of the CPU. We expect to have this system run-
ning on the UNIX 4.2 BSD SUN software, too. The base system
on the DUAL, including the interpreter, reader, bignums,
fasl, works; the compiler is being perfected.
5. Other Lisps
We now have, in-house tried 4 (other) VAX UNIX lisp
systems: YLISP, Interlisp, PSL, and VLISP. We know that
Interlisp can run also on VMS using the Eunice package.
Interested parties can contact David Dyer at USC-ISI. There
is also a version of lisp which runs on VMS only, namely
NIL, from MIT, which appears to be undergoing limited dis-
tribution. Two other lisps under development under UNIX are
Yale's answer to NIL, namely "T", and Common Lisp, from CMU
and friends.
Counting Franz, that makes 7 lisp systems for the VAX
computer line. Not counting variants on 2 operating systems.
A Paen to standardization.
Dick Gabriel states some useful principles for com-
parisons in the conference record of the 1982 ACM Lisp and
Functional Programming Conference, which was held in August.
We understand he now has a collection of some 18 programs
which he is continuing to time on various systems.
6. Work in Progress
6.1. BITGRAPH SUN AED-512
Greg Foster at UCB is working on raster-graphics sup-
port in Franz for the 800 by 1000 b/w raster displays of the
BBN Bitgraph and/or the SUN Workstation, and possibly the
color 512 by 512 AED system. We are probably going to han-
dle mice and Bitpad (stylus) input for pointing. There are
lots of projects we hear about with similar systems, e.g.
just recently from the University of Maryland, using UNIX
and multiplexed files for window management of a 68000-based
home-grown workstation.
6.2. RISC-LISP
Yes, Reduced Instruction Set Computer fans, who else
but UCB would be so bold... Carl Ponder is examining the
issues involved in constructing a fast lisp interpreter and
compiler for the RISC architecture. You see, we have all
these chips...
7. Work Contemplated
7.1. Fast Number Compiler
Undergraduate Jeff Cohen at Berkeley is starting to
look at this. There are several industrial concerns that
have expressed interest in using such a system, but expected
it to be subsidized by someone else.
7.2. IBM Franz
Even more nibbles on this one, but not yet.
8. Business News
8.1. Eunice SOLD
Some of you may have heard that the Eunice software
package was sold by SRI to the Wollongong Group, a UNIX sup-
port group in Palo Alto. Prices range from $2k (educa-
tional) to $5k (commercial). Naturally this package is of
interest beyond the availability of Franz Lisp. We have not
compared this product to other similar ones, but we know
that TWG has been distributing a working Franz opus 38.
As far as alternatives to Eunice, we are aware of a
system developed at Rice University, and another by Human
Computing Resources (HCR) in Toronto. We have not
evaluated either of these.
8.2. MACSYMA SOLD
MIT has sold exclusive rights to MACSYMA, a large alge-
braic manipulation system, to Symbolics, Inc. of Cambridge
Mass. This package runs in Franz Lisp, (among other Lisps)
We hope that soon it will again be available to educational
institutions with VAX systems either from us or Symbolics,
at a nominal charge. We understand that commercial licenses
(from Symbolics) for versions of MACSYMA on PDP-10s, Lisp
Machines, etc. will distributed at non-nominal prices and
offered with maintenance contracts.