On 13 June 2013 02:43, Kelly Dean <kellydeanch@yahoo.com> wrote:

> Michael Heerdegen wrote:
> >I often do C-y M-y M-y ... like most of us, I think.
> >Currently, if you `undo' after that, each of the M-y's are undone
> >separately.
> [...]
> >I would suggest that undo simply undoes until before the first yank.
>
> I agree. I think if you put an undo boundary before yank, and before any
> change besides yank-pop that follows yank, but don't put one before
> yank-pop, then when undoing and redoing, it would act like the text
> insertion done by the last yank-pop was the only thing that happened.
>
> Andreas Schwab wrote:
> >You can always redo if you overshot.
>
> Yes, but having to undo through all the yank-pop events in the first place
> is annoying when you only did them for the purpose of reviewing the last
> few items on the ring, and didn't intend to leave the text inserted. And if
> you overshoot while doing yank-pop, you don't need undo to go back anyway,
> since you can use a negative arg with yank-pop (which I do so often that I
> have a yank-pop-rev command that does that, with a dedicated key chord
> bound to it)


Uh huh. IMO there should be such a keychord by default and it should be
short and advertised.