Re: New undo element (fun . args)

[Stefan Monnier <monnier@iro.umontreal.ca>]

> A _lot_ of people are annoyed at the way `undo' also redos things.
> For that reason, a package redo.el exists, and people get told about
> it frequently on Emacs lists and groups.

> I consider it likely that with the new changes, this package will no
> longer work.

It's possible of course, I haven't looked at how it's written, but I find
it unlikely.  The most obvious way to implement it doesn't need to look at
the undo elements.

> I also have detected the variable undo-no-redo (not customizable, though)
> which wouls appear from its name and description to solve one half of the
> problem,

The command is called `undo-only' and I had originally bound it to
C-x U (i.e. a stronger version of C-x u since it skips over matching
undo-redo pairs in the undo-list), but Richard reminded me that we do
not like to bind different commands to uppercase and lowercase commands and
I haven't bothered to try and find another binding for it.

> the remaining half being that an actual redo command working in
> the opposite direction would be required.

C-f C-/ C-/ C-/ ...

Maybe we should arrange that calling `undo' after `undo-only' does not
continue undoing but switches to redo instead, so that redo.el would amount
to (defalias 'redo-undo 'undo-only) (defalias 'redo-redo 'undo).

> Since we are likely to break redo.el, wouldn't it be sensible to
> provide the functionality from within Emacs once and for all?

I find redo.el to remove functionality rather than to provide it.
But I did go through the trouble to implement undo-only.


        Stefan