🚀 go-pugleaf

In article <588@svax.cs.cornell.edu>, Matthew Belmonte 
(..!decvax!duke!duknbsr!mkb writes):

> recently i spent an evening going through mounds & mounds of Symbolics's
> Zmacs code in an eventually successful attempt to find a way to bind a

> You already have about 10 good manuals for
> the machine.  How about "11:  System Software Implementation"?
> ("Oh, yes, it's another $40000 for the manuals.":-)

Welcome to the Lisp Machine Hacker's club, Matt. I suppose that the problem
with writing a manual on the software implementation of Zmacs, windows, etc.
is that Symbolics would have to somewhat support that implementation
across new releases. that means lotso money and time (i.e. resources).

If you continue using the LM as a development tool,
you'll undoubtedly be digging into the window system, as well as the
menu system, mouse control, and even more into customizing Zmacs. Without 
system source code this probably would not be possible. The fact is, it's
a mess under there. I suspect that by not supporting manuals on the
underlying datastructures, Symbolics will be able to "easily" clean some
of this stuff up - as time marches on - and not have to worry about
customers whining about changes. It's interesting to note how the new
release (Genera) attempts to raise the level even higher (i.e. by using
Frame-up for window/pane design, SemantiCue for mouse control, and
Dynamic Windows for output control). 

Because of the amount of customizing we've done here (over 1.5 years worth),
I hope the foundation is the same as with release 6. Of course they 
say it is... (except for CL compatability changes), but one never 
really knows, eh?

Christopher Rosebrugh, GTE Labs, Waltham, Mass.  ...!harvard!bunny!cpr0