On Thu, Oct 02, 2003 at 12:16:18PM +0200, Lars Gullik Bjønnes wrote: > Andre Poenitz <[EMAIL PROTECTED]> writes: > > | On Thu, Oct 02, 2003 at 11:29:08AM +0200, Lars Gullik Bjønnes wrote: > >> Angus Leeming <[EMAIL PROTECTED]> writes: > >> > >> | The trouble as I see it is that the real problems in the 1.4.x tree stem > >> | from the kernel being in a state of transition from old design to new. My > >> | reading of discussions from those who know is that fixing some (major) > >> | bugs _requires_ this transition to be pushed on. > >> > >> I do not belive this. What I think is the case is that those in > >> "charge" of the transition are unwilling > > > | 'unwilling' is too strong a word I suppose. > > Perhaps. > > >> to work on making a intermediate state stable. > >> > >> (anyhow I see noone moving the transition forward...) > > > | I've put out a 'small steps' plan where the steps are of a size where > | everyone can contribute. > > This was for a unified cursor only, right?
The 'unified cursor' (at least the first part: deep cursor for the outside world, I don't speak of 'outside/math unified cursor' in 1.4 anymore) is the same as 'getting rid of inset locking'. Inset locking is bad for several reasons, cursor movement is one of them. Furthermore, I have the gut feeling that the 'uninitialized' LyXCursor crashs play on the same meadow: We access the LyXCursor of some LyXText before it is in a decent state. Formerly, this was no problem, as the Paragraph * = 0 was set on creation and 'handled' properly in the 'too early calls' by simply returning early. In order to initialize the iterator used nowadays properly we'd need the Parlist, obtained by LyXText::ownerParagraphs, which bombs in an uninitialized LyXText. This is not saying the ParagraphList::iterators are a bad thing, it's just saying we are not able to brush over problems as we were before. Andre'
