On Fri, Apr 28, 2006 at 05:00:06PM +0200, Jean-Marc Lasgouttes wrote: > >>>>> "Martin" == Martin Vermeer <[EMAIL PROTECTED]> writes: > > >> Because we had a completely different update system? > > Martin> Yes, that's what I suspect too. So we must get this working > Martin> from scratch then. > > Martin> Could lyxfunc be made sensitive to LyXAction flags? Would that > Martin> be part of a solution? > > Maybe, but what flag? Or add a fitcursor() member to DispatchResult?
That's pretty much the 'invalidate small chunks, then update' approach. > A LyXAction flag would be useful to handle this new member easily in > the easy cases (like we do with NoUpdate). I am not too fond of the LyXAction flags as this requires a knowledge 'outside the handler' about what happens in the implementation. Moreover, it could be that some branch of the handler causes a completely different invalidataion pattern. Andre'
