HA ! That's a bloody joke. This doesn't work. I have absolutely no idea why, of course. This update stuff is entirely beyond understanding. Just dealing with the bv->updateInset() vs. bv->update() vs. bv->update(blah) vs. inset->update() vs. inset->updateInsetInInset() vs lyxtext->update() vs. lyxtext->updateInset() [1] isn't too bad, but the semi-recursive nature and delayed insettext state update thing makes it impossible. I've spent 2 solid hours trying to work out the correct way of forcing an update of a minipage, no go. Pretty pissed off ...
I would have been *really* surprised if you would have figured out a better way in one day, do you know how long I worked on this stuff to make it work as it is now.
The only way I see to fix the whole lot is that we would have to update the deepest insets first. I'm pretty sure there is a good algoritmus for this and if one goes and concentrates on the problem I'm pretty sure we come up with something working. As I told you we have to update the deepest insets first and then from there come up. Obviously at each level we can have more then just 1 inset and that is what complicates all of it, but as I said I'm pretty confident that there is a better way to solve this as the actual one which (I admit it) is pretty hard to understand and far away from the best way to do it.
I've done this anyway, since it doesn't make things worse. I'm committing the lot below tomorrow.
Ok you did it as I expected. Defined a fixed font for the empty tabular cell font.
What I don't like in all this is the removal of the "mark_dirty" flag how do you do now to set the dirty flag on the buffer? (I admit I had only a fast look at your patch).
Jug
-- -._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._ Dr. Jürgen Vigna E-Mail: [EMAIL PROTECTED] Mitterstrich 151/A I-39050 Steinegg Web: http://www.lyx.org/~jug -._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._