Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: Notesfiles $Revision: 1.6.2.17 $; site smu.UUCP
Path: utzoo!linus!philabs!cmcl2!seismo!harvard!talcott!panda!genrad!decvax!harpo!whuxlm!whuxl!houxm!ihnp4!inuxc!pur-ee!uiucdcs!smu!pedz
From: pedz@smu.UUCP
Newsgroups: net.emacs
Subject: Re: Mlisp Function Inquiry
Message-ID: <12400011@smu.UUCP>
Date: Wed, 20-Feb-85 20:27:00 EST
Article-I.D.: smu.12400011
Posted: Wed Feb 20 20:27:00 1985
Date-Received: Sun, 24-Feb-85 02:51:08 EST
References: <382@isrnix.UUCP>
Lines: 19
Nf-ID: #R:isrnix:-38200:smu:12400011:000:1009
Nf-From: smu!pedz    Feb 20 19:27:00 1985


I think of a "command" as something that starts at the root of
the keymap tree and eventually terminates a function being
called.  Added to this, I would exclude some commands from the
definition.  In fact, I would make a mlisp command which has the
effect of saying "don't count me as a command".  (A rule forcing
that this command be the first statement in the mlisp function
would be reasonable.  I would definately make this command act
only if it was in the function called directly by the keymap.
Thus if a function which wishes to be counted as a command,
called a second function which did not, the first function would
still be remembered).  This could be done by simply keeping a
two entry queue of commands and when a command was called, it
would get put into the queue.  If the function did not want to
be counted as a command, it could call the "forget" function
which would backup the queue.  I really have no idea of what
these ideas involve.  Perhaps someone else could comment futher.

Perry