Richard Heck wrote: > 5. Charstyle drawing shouldn't mess up line breaking the way it does > now. The insets get drawn as if they are single characters, so you end > up with: > This is some > text. This is some > text. > [http://really.long.url.that.messes.up.the.line.breaking]. > This is not good. CharStyle insets should be drawable across lines in > the natural way.
Just a quick thought. It seems to me 1) very hard to achieve the nice line-breaking behaviour we have now by having insets as entities that draw themselves. I think the 3 boxes scheme is very hard to achieve and ultimately sort of silly. 2) very hard to achieve cursor movement in a reasonable way with inset handling lfuns by themselves. All this may indicate that we are really trying to force things. So my proposal is to have charstyle as half-insets. The data structure could be that of insets (so latex export could work easily in a recursion scheme), but screen representation and cursor handling could be taken care of by the containing insettext (for instance, with a dociterator-like interface it could have access to the raw stream of chars to make rowbreaking with the very same current code). It's not that I've thought about all implications in detail, and I'm may be plain wrong, this is just sort of my gut feeling about it. A/
