Re: [CVS] f7, f8 bound..

[Per Abrahamsen <abraham@dina.kvl.dk>]

"Robert J. Chassell" <bob@rattlesnake.com> writes:

> Hmmm.... I occasionally use Customize (mainly for faces, which I do
> really not understand) and I use setq, and also, I hand edit the
> expressions that Customize inserts in my .emacs file.

With either Alex or my code, what is saved to the .emacs file will
look nothing like what you would write by hand.  You can still edit
it, though.

> Please, arrange that customize and everything else work with just
> *one* variable.  Not two.  Otherwise, you will make maintenance much
> harder.  Thank you.

I believe maintenance will be much harder with one variable that tries
to do everything, rather than two variables with clearly identified
purposes.

Example:

foo-mode.el:  Defines foo-mode-map.
foo-extra.el: Defines some new bindings for foo-mode-map.

With Alex current code, the foo-extra bindings will be removed each
time the user make any changes to his own bindings.

We could change Alex' code to remember the old user bindings, but then
the result will depend on whether the user bindings were were made
before or after foo-extra.el was loaded, and if the standard bindings
change in foo-mode.el, removing the user bindings will give you the
old binding, with the risk of weird errors.

We may be able to overcome these problems, or choose to ignore them,
but in either case I believe the end result will be much more complex
than simply having two keymaps, one for customize, and one for Lisp.

We don't need the extra non-keymap variable, Alex code use a symbol
property instead, which should work just as well.  Only customize have
an interest in that value, so there is no problem hiding it that way.

We _can_ also use a property for the customize keymap, but since that
is the map the mode need to enable (with use-local-mode), it will have
to know about it.