On Thu, 2005-11-03 at 14:09 +0100, Jean-Marc Lasgouttes wrote: > >>>>> "Martin" == Martin Vermeer <[EMAIL PROTECTED]> writes: > > Martin> On Thu, Nov 03, 2005 at 09:06:17AM +0100, Helge Hafting wrote: > >> The bugzilla entry is now updated with a test document and this > >> crash recipe. It trips the assert everytime for me: > > Martin> Here's the fix. > > I do not think it is the right fix. AFAIK a lot of code relies on > the fact that selectionBegin and selectionEnd are in the same inset. > > Isn't it possible to ensure that? > > The simplified crash case is > > - create a footnote with "text" in it > > - place the cursor in front of text (inside the text inset) and select > to the right (mouse or keyboard) until the cursor is outside the > inset and the whole footnote is selected. > > - do any font change (e.g. C-b) ==> crash
No it doesn't. Not with this fix. > I suspect that the last step could be many different things... Yes. > A possibility could be to change selectionBegin/End to return the > right cursors. However, the display of the selection is correct, so I > presume there is some code somewhere that handles that case at least > for selection painting. Where is it? Good question. Alternatively, redefine the != operator for dociterators to take depth difference into account. > JMarc - Martin
signature.asc
Description: This is a digitally signed message part
