Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84 exptools; site ihwpt.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxn!ihnp4!ihwpt!warren
From: warren@ihwpt.UUCP (warren montgomery)
Newsgroups: net.emacs
Subject: Re: Montgomery's Emacs
Message-ID: <615@ihwpt.UUCP>
Date: Tue, 17-Dec-85 12:56:57 EST
Article-I.D.: ihwpt.615
Posted: Tue Dec 17 12:56:57 1985
Date-Received: Thu, 19-Dec-85 05:16:18 EST
References: <733@mit-eddie.UUCP> <11600007@ccvaxa>
Organization: AT&T Bell Laboratories
Lines: 59

My emacs implementation was begun in 1979, after having left MIT.  I
made it freely available to people INSIDE of Bell Labs, and it was
widely used.  It was never officially "released" from Bell Labs.  At
the time, Bell Labs had a policy of not copyrighting software, but
instead trying to protect it by keeping it secret within the
company, thus no code was ever marked with copyright notices. 
Unfortunately, several copies did get out during that time, mainly
due to people who left Bell Labs to return to school or gave copies
to friends.  When Zimmerman modified one of those copies as the
original basis for CCA emacs, AT&T and CCA had a prolonged debate
over it.  Eventually the matter was resolved when Zimmerman replaced
the last of my code, and AT&T has tried to be more careful with it
ever since.  

To summarize, I don't think my employer considers any of the escaped
copies of my emacs to be public domain, though there may be legal
grounds to contest that.  (I'm not a lawyer, so don't take my word
for it).

As for what's good or bad about my implementation, here is some
information:

Mine is probably the least extensive of the unix emacses, as it has
never been even close to a full time job for me.  I emulated the
emacs that I had come to know and love at MIT, as of 1978, so it
diverged a little earlier than the others.  It is certainly the
smallest in code and data of the various implementations, and runs
quite well on split I&D pdp-11's, memory starved PC's, and other
small machines (though I doubt it would run on an 11 with only 64K
total I&D any more).  The last time I saw any timing comparisons, it
ran much faster than Gosling's, though that was years ago.  The
editor is written entirely in C, which is an advantage in speed and
a disadvantage in flexibility.  It is reasonably straight-forward
code, which is an advantage to anyone who wants to hack it up, but
as a result it doesn't optimize odd cases at all.  It also depends
relatively little on it's environment, relying only on the section
II system calls, which has made it easy to port.

It now comes with an extension language that is lisp-like, and is
compiled to teco-like gobbledygook, and provides a good deal of
flexibility.  You are still limited in changing the behavior of the
stuff hard coded in C. It has a solution to the flow control problem
that in most cases allows it to work fine with networks or terminals
that need ^S/^Q for flow control and still allow you to type ^S and
^Q as commands.

As for the pricing, AT&T's pricing structure is different from
industry norms.  The single user price listed ($950, or something
like that) is for a single purchaser to do anything they want with
it internally.  Thus if you have a hundred vaxen and want to run it
on all, you pay just the one fee, making it a good deal for
organizations that want a lot of copies.  A major objective of the
Toolchest is to make software available to people who will
re-package and support it.  For a one time fee of under $10K, you
can buy the right to repackage and sell it however you want.  

-- 

Warren Montgomery,  ihesa!warren