Re: [STUMP] Functions vs Commands?

[Eric Abrahamsen]

"J. David Smith" <address@hidden> writes:

> Hullo,
>
> I'm sure most of you are aware of the fact that (define-key ...)
> cannot directly take a function as its third parameter. Instead, it
> is given a string containing the command name. For example, this:
>
> (define-key *root-map* (kbd "f") "firefox")
>
> versus this:
>
> (define-key *root-map* (kbd "f") #'firefox)
>
> This is doubly confusing because (defcommand ...) wraps (defun ...),
> so (defcommand firefox (&optional (new-window nil)) () ....) produces
> a function (firefox ...). In order to make two keys (f and C-f), I
> can't simply do:
>
> (define-key *root-map* (kbd "f") #'firefox)
> (define-key *root-map* (kbd "C-f") (lambda () (firefox t)))
>
> I have to define two commands:
>
> (defcommand firefox ...) and (defcommand new-firefox ...).
>
> as well as two keys.
>
> So I suppose my question can be summed up like this:
>
> What is the reasoning for using strings to reference functions like
> this?
>
>   -- J David Smith

I can't really explain the reasoning, not having been in on it, but I do
know the system was put in place as a kind of echo of Emacs' distinction
between functions and interactive commands.

Luckily, your problem is fairly easy to solve -- you can pass command
arguments along with the command name as part of the same string. So
try:

(defcommand ff (&optional (new-window nil)) (:y-or-n)
            (if new-window
              (message "new window!")
              (message "no new window!")))

(define-key *root-map* (kbd "z") "ff")
(define-key *root-map* (kbd "Z") "ff t")

There's no "boolean" type, but any string will do.

If you're using StumpWM from git, update to something recent and check
out the manual -- there's a new section on this.

Hope that helps,
Eric