On Thu, Feb 05, 2004 at 01:13:12PM +0100, Alfredo Braunstein wrote: > > I'd think we should settle here for a clear scheme, i.e. either absolute > > or screen-relative. > > screen-relative is out of the question: it's clearly worse than absolute. > The only things that can use screen-relative coords are on early stages of > the dispatch process (say that workareadispatch could use it for mouse > gestures or whatever) and on last steps of the draw process (for obvious > reasons). > > Except that two cases we can take the screen business out and work with doc > absolute coords everywhare I think.
Fine with me. > >> - inset::dispatch and lyxtext::dispatch take absolute coords (ok) > >> > >> - the cached x,y position in insets is screen-relative (bad) > > > > I can't really make up my mind with all this coordinates business. > > 'Everything absolute' would be nice, but we can't update the caches all > > the time... > > Why would we need to update them all the time? we update them at draw time > exactly as now, and the only ones who have correct info are the ones shown. > Again, I think that absolute beats screen-relative in every aspect. Like 'scroll jumping' in the middle of nowhere. But you are right, screen-relative make that even worse. > (putting absolute coords in the insettext cache should be automatic at this > point) > > For internal things LyXText-relative, or par-relative is fine I think (as > long as things remain somewhat private). Which 'internal things' do you mean? Andre'
