On Sat, Jan 17, 2009 at 10:00:57PM +0100, Stefan Schimanski wrote: > > Am 17.01.2009 um 20:33 schrieb Andre Poenitz: > >> On Sat, Jan 17, 2009 at 06:40:05PM +0100, Stefan Schimanski wrote: >>> Did you check what exactly needs so long in updateMacros? It >>> essentially >>> just looks at the insets of each paragraph. >> >> Which is still a bit. >> >>> If there is no macro >>> definition, nothing is done. Otherwise, the macro table of the >>> buffer is >>> updated. I don't see why updateMacros needs so much more time than >>> the >>> updateLabels mechanism. >> >> Why can't that be triggered 'manually' on buffer load and on changing >> a macro definition? > > Because the macro position is not constant between these points in time. > Therefore the lookup tables would be invalid then.
But that means that we'd need something more robust than a DocIterator as pointers. One more reason for "registered iterators"... Andre'
