Am 31.05.2007 um 21:57 schrieb Dov Feldstern:
Debugging right now. But still no idea... Somehow the computeTable function accesses an invalid character.Dov Feldstern wrote:Dov Feldstern wrote:I'm getting a crash now... :( I still don't know which of the above (if any) is the culprit, I'm trying one-by-one to take them off my patch stack, but the compile-test cycle is taking a while.As always, Stefan, you are right :) .I applied your patch (what threw me off was what you said about including another patch --- which BTW, I would still commit separately). Then it didn't compile --- there's a semicolon missing on line 347 of the patch file, or at line 279(?) of bufferview_funcs.cpp.But now it works, and like a charm: the selection patch you sent yesterday works correctly in the case described below now. And also in the case of the problem I reported yesterday in bugzilla of bug 3550. So you're definitely doing something right ;) .So all of these patches:*) http://permalink.gmane.org/gmane.editors.lyx.devel/85956 (fix for bug 3754: cursor movement at RTL-LTR boundary) *) http://permalink.gmane.org/gmane.editors.lyx.devel/86192 (this fix, for issue (3) in bug 3551 --- except note that this also includes the above patch, they should be separated, as that one is independent of this one) *) http://permalink.gmane.org/gmane.editors.lyx.devel/86105 (fix for bug 3550: selection in bidi)are OK with me, and should be applied (in the above order).Anyhow, to reproduce:1) type a long long string with no space, until it breaks to the next line. 2) from the last position in the paragraph (it only works from there) start backspacing. When you try to delete the last remaining character on the line -- kaboom!Attached is the trace; looks familiar...The crash occurs in a clean svn, so these patches are still OK by me.I'm working on the crash, Stefan would probably be able to solve it much more quickly...
Stefan
PGP.sig
Description: Signierter Teil der Nachricht
