Re: undo bug?

[martin rudalics]

> The record-point after it wouldn't make anny difference, would it?

It's needed for getting there after a redo.

> Hmm... unless undoing a `record-point' would forcefully add another
> record-point entry i nthe undo-list... I guess that might work.

Any `record-point' would simply have to preserve the symmetry of

UB RPB buffer-changes RPA UB RPB buffer-changes RPA UB

where UB is the undo-boundary, RPB and RPA the `record-point's before
and after the buffer-changes, and the buffer-changes would not contain
any undo-boundaries or point-recordings.

> Still, I'm wondering what will be the effect of such a thing on the
> length of the undo-list.  It's a pretty minor issue, as far as I can
> tell, so it had better not have any noticeable downside.

We could remove certain text-properties from undo-lists (mine are
usually full of 'fontified properties in deleted text).