Re: [Emacs-diffs] trunk r118062: * lisp/icomplete.el: Move Iswitchb autoload here. Much simpler.

Re: [Emacs-diffs] trunk r118062: * lisp/icomplete.el: Move Iswitchb autoload here. Much simpler.

[Glenn Morris]

Existing customization of iswitchb-mode to non-nil will no longer be
respected following your change, so you might want to bring the rest of
this back:

http://lists.gnu.org/archive/html/emacs-diffs/2014-03/msg00325.html

The whole point of this was to be re-usable the next time this was
needed, without hand-written autoloads; but whatever, I imagine you'll
say there won't be a next time.

(The reason this got more complicated was bugs in autoload generation.)

Re: [Emacs-diffs] trunk r118062: * lisp/icomplete.el: Move Iswitchb autoload here. Much simpler.

[Stefan Monnier]

> Existing customization of iswitchb-mode to non-nil will no longer be
> respected following your change,

We'll see how many users scream when we start the pretest for 25.1.
I think it can be considered as one more step in the "progressive
removal of iswitchb":
- in 24.4 it is marked obsolete (and causes a corresponding `message'
  when loading) but otherwise still works as before.
- in 25.1 it is still autoloaded as a function, but not as a variable or
  custom setting.
- in 25.2/26.1 it might get fully obsoleted (i.e. not autoloaded at all
  any more).

> The whole point of this was to be re-usable the next time this was
> needed, without hand-written autoloads; but whatever, I imagine you'll
> say there won't be a next time.

Well, yes, I think it was a mistake for me to accept moving iswitchb.el
to obsolete.


        Stefan

RE: [Emacs-diffs] trunk r118062: * lisp/icomplete.el: Move Iswitchb autoload here. Much simpler.

[Drew Adams]

> Well, yes, I think it was a mistake for me to accept moving
> iswitchb.el to obsolete.

Then why not move it back to normal status (not obsolete)?