At 05:21 PM 3/24/01 -0500, Dom Lachowicz wrote:
>With some further digging, the call boils down to
>px_ChangeHistory::isDirty(). Basically, we say that the doc is dirty if the
>undo position at last save != the current undo position.
>
>To see what I mean, type "foo", then save. then type some more and hit undo.
>Everything gets removed.
>
>This might be the correct way to do Undo, but it's screwing up marking stuff
>as dirty. I think that I can nail this bug. More to follow.
The original intent was that as you undo/redo, there's a single state
(corresponding to the last save) which is *not* dirty. Everything else is.
Is there a coalescing problem? If so, perhaps the solution is to add
special case logic to break the undo coalescing chain at the save point.
It's been quite a while since I looked at that portion of the PT code, but
I'm guessing it shouldn't be too hard to localize such a change.
Paul
