Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!sri-unix!hplabs!decwrl!labrea!kestrel!king
From: king@kestrel.ARPA (Dick King)
Newsgroups: comp.emacs
Subject: comparison shopping
Message-ID: <14582@kestrel.ARPA>
Date: Mon, 17-Nov-86 14:42:42 EST
Article-I.D.: kestrel.14582
Posted: Mon Nov 17 14:42:42 1986
Date-Received: Mon, 17-Nov-86 22:31:26 EST
Organization: Kestrel Institute, Palo Alto, CA
Lines: 63


   From: mg@unirot.UUCP (Mike Gallaher)
   Newsgroups: comp.emacs
   Date: 14 Nov 86 20:16:43 GMT
   Summary: don't knock it 'til you've tried it

   That message was a (heavily-encoded, sorry, but not sarcastic) response to
   statements such as "Gnuemacs is the best Unix Emacs around".  I certainly
   do pay attention to constructive well-researched comparisons, but these
   were not couched as such, so I interpreted them as being religiously
   motivated.  I don't take them seriously, any more than I am serious myself
   when I say things like "Unipress Emacs is the greatest text editor in the
   solar system" (which it is).  (After working on it for more than two years,
   naturally I think so!)  But for those who don't know better, such comments
   could give Gosling Emacs a poor reputation that I don't think it deserves.
   I want to encourage people to try the choices themselves, and then decide.

We did just that.  At our site (30 people, using editors mostly for
TeX files (and editor customization files :-) )) we made both
available for six months.  At the end of the six months, the
last-reference date on Unipress emacs was three months in the past,
and when it was removed from the system nobody noticed.  

This experiment took place in 1985.

   For the record, I am not anti-Gnuemacs; it's got some spiffy things in it,
   although RMS and I will probably never agree on various design issues.

I, for one, would show Unipress EMACS a lot more respect if it had
conses.  

As an example, I got annoyed that UNIX doesn't have file version
numbers, so I hacked up version numbering in GNUmacs in 2 hours.  I
then tried to do the same thing in Unipress EMACS, after having had
"practice", and gave up after about 8 hours.

I submit this, neither to be argumentative, nor to "run down"
unipress; I submit this to try to send you a message, stronger than
the one I used to send every couple of months by calling Unipress,
that this one feature is important.  No, we're not going to spend our
days hacking .ml or .el files, but the rare occasions when we notice a
nagging splinter in our sides and want to fix it, certain things are
vital!  Note that I did NOT harangue you for things like the rather
baroque argument passing method.  It is annoying but not fundamental,
and to change it now would obselete code.

Mr. Gallaher, PLEASE GIVE US AN EXTENDED DATA STRUCTURE OF SOME SORT.
Conses would be easier and what we're used to.

(I still use Unipress EMACS for my PC.  Before you respond something
like "see, conses make GNUmacs too bulky" I will point out that the
LISP system I wrote for the PDP-11 devoted only about 2K bytes to
garbage collection, and you could make that swappable.)

   Btw, the GOSDIFFS file in the GnuEmacs distribution (whatever the last
   17.xx was) is out of date, but I think someone besides me should fix it :-).

   Mike Gallaher
   Emacs Hacker Boss
   Unipress Software

Please counterflame to me as well as to the group.

-dick