Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!philabs!cmcl2!seismo!hao!hplabs!sri-unix!Info-IBMPC@USC-ISIB.ARPA
From: Info-IBMPC@USC-ISIB.ARPA
Newsgroups: net.micro.pc
Subject: none
Message-ID: <628@sri-arpa.UUCP>
Date: Thu, 2-Aug-84 05:07:47 EDT
Article-I.D.: sri-arpa.628
Posted: Thu Aug  2 05:07:47 1984
Date-Received: Sat, 4-Aug-84 08:28:53 EDT
Lines: 90

From:  Info-IBMPC Digest 

29-Jul-84 01:02:40-PDT,4501;000000000001
Date: 29 Jul 1984 01:02:39 PDT
Subject: Unipress Gosling EMACS
From: Billy 
To: info-ibmpc@USC-ISIB.ARPA

ISI has recently acquired a beta test version of Gosling EMACS for
the IBM-PC. For the last few years I have been using MINCE from
Mark of the Unicorn on my PC and EMACS on Tops-20. I assume readers
are familiar with the various dialects of EMACS and so will
concern myself only with details of implementation in this article.

Gosling EMACS from Unipress is a complete implementation. The .EXE
file alone is more than 275K bytes. The libraries fill up another
640K disk and the help database (which isn't shipped in this version)
is rumored to take up nearly 2 megabytes. While EMACS will run in
384K I wouldn't recommend running it in less than 512K.

The entire program is written in Lattice C and a source license is
available for a reasonable $999 price. I think the normal object
license is around $350 or $375.

It would be unfair to review this product based on the beta test version
as there are a number of known problems that should be worked out
soon.

EMACS takes about a minute to load from a floppy disk. I am told
this will be speeded up in the future. I believe much of the time
is taken in a MLISP compilation that will be eliminated. (MLISP is
the internal implementation language for Gosling EMACS)

Once EMACS is loaded one can push to an inferior EXEC and run
compilations or whatever. I have not tried the "compile" command,
but theoreticly one can compile a set of programs in an inferior
exec and monitor errors in separate EMACS buffers. I have used the
"push to inferior exec" command and run compilations of programs by
hand and instantly returned to the memory resident EMACS with my
source files loaded and ready for a reedit. This feature is real
handy and speeds things up immensely but has its perils. You need
buckets of memory as EMACS with reasonably full buffers takes up
about 300K, and as the Pascal Compiler isn't exactly miserly I take
full advantage of the 1.25 megabytes of memory on my PC.

This version of EMACS will probably be of use to programmers
because of this feature, but the IBM-PC's lack of memory protection
really bytes you here. The folks at Unipress warned us in the
manual that badly behaved programs running while EMACS is loaded in
memory could wipe out EMACS internal structures and EMACS could
then proceed to trash the disk.

While memory protection would be nice, I am sure this sort of error
could be avoided by protecting critical data structures and code
with checksums.

Given the size and complexity of the program and slowness of the PC
this implementation of Gosling EMACS performs very well. Running it
on a PC is significantly better than running on our VAX-780 against
a half dozen users in a typical daytime load. I am sure the program
will get faster as it is honed, and an upgrade to an 8086 or one
of the newer Intel processors would make this direct screen implementation
faster than a 9600 baud connection to a stand alone VAX.

I am suffering a bit from culture shock. I have never even configured
an init file in Tops-20 EMACS much less dealt with the MLISP code of a
Gosling EMACS. Dale Chase here at ISI is working on adapting MLISP
mail reading libraries so that they are suitable for the PC environment.
I would like to consider MLISP as a valid language for submission of 
programs to the INFO-IBMPC library now that there is a Gosling EMACS
available for the PC, but I am utterly incompetent to administer such
a library. Perhaps someone from CMU who is more familiar with MLISP
could act as a net source for MLISP libraries appropriate to the PC.
Are there such mailing lists already?

I'd love to throw away my copy of Mince, but Unipress isn't quite
there yet and I am back to using Mince for most of my editing. I am
sure the real release version will be more customized to the PC and
clear up the obvious annoying bugs. Unipress does offer a smaller
version of EMACS which does not include MLISP capability, but that
won't be ready for the PC for a fairly long time. While this
current version of the program leaves a lot to be desired, things
look good for the future as the program has a lot of potential and
the people at Unipress have been very pleasant to deal with and
seem aware of the problems and are interested enough in our
opinions to send us a beta test version.


-------
-------