On Sun, Aug 11, 2002 at 09:25:44PM +0300, Martin Vermeer wrote: > There are pluses and minuses about this patch, though. > > 1) updateCounter is called from a number of locations with different row > arguments (the second arg). I don't quite overview all the implications > of getting rid of this argument (which we can with this patch, a plus I > think). Please give it some good testing.
If you are going to pass over the whole buffer anyway (which is not un-sensible per se) there is exactly one occurence when it _has_ to be finished: Just before redraw. So one option would be not to update the counters only there. If we do not run into performance problems this might be all it takes. > 2) updating all counters instead of just the remaining ones requires a > little bit more resources. A worthwhile trade-off? (We do a lot less > copying of counters now, with buffer counters.) I think this was a > premature optimization that can go now. Please check this, but I have a gut feeling that you might be right. > 3) Did anybody else notice that > text.C/text2.C/lyxtext.h/paragraph.[Ch]/buffer.[Ch] is a sordid mess? > I don't like tiptoeing through minefields... Nobody noticed this lately as everybody is avoiding to loooks at it ;-} > What shall I do with this? Re-do it? Ask Lars to get his test tree working? Andre' -- Those who desire to give up Freedom in order to gain Security, will not have, nor do they deserve, either one. (T. Jefferson)
