Am 18.05.2011 um 08:36 schrieb Peter Kümmel: > On 18.05.2011 07:33, Stephan Witt wrote: >> Am 17.05.2011 um 23:21 schrieb Murat Yildizoglu: >> >>> I think I have found the other conflict: I have recently bought Forever >>> save 2 in a bundle and installed it. Since Lyx is my main editor, I have >>> configured it for taking snapshots of Lyx 2 documents and I think it >>> crashes Lyx as soon as it tries to plug itself on it. I have suppressed Lyx >>> from the list of programs watched by Forever save and Lyx stopped crashing. >> >> That's good news. At least there is a work around. >> >>> This makes conflicts with two very useful tools... The lost of Forever save >>> is only for Lyx, since it is enough to make it ignore Lyx, but the problem >>> with Divvy seems more general. But I will test it now again, since it >>> should be OK as long as I do not call Divvy for resizing the Lyx Window. >>> OK, given my first experiments, this seems to work. Lyx has been launched >>> without crashing. If I meet another crash, I will inform you. >>> >>> Any idea about the conflict with ForeverSave 2? I will ask them the >>> question too. >> >> To ask that people would be good. One problem with these tools I have is >> that they are not free. >> After using Divvy some times I lost the functionality and cannot test it >> anymore. The same holds >> true for ForeverSave - ok, only some Euro but I cannot buy and install every >> tool out there... >> >> @Peter: the crash is attached to the first mail in this thread and is >> similar to the one mentioned in ticket 7365. >> http://www.lyx.org/trac/ticket/7365 >> >> I could reproduce it at my machine - it's an endless recursion, you can see >> only the 1st 511 stack frames. >> I had appr. 32000 of them in the debugger when it crashed and there is no >> LyX code involved. I have to search for it >> again. > > Looks hard to find. Good luck. Maybe there is a hint in Qt's ticket system or > qt/src/core/kernel history.
Yes, I've to use google and diff the 4.6 and 4.7 sources. Of course it's possible that we trigger a bug fixed in 4.7 only. > >> >> I suspect the session manager code of Qt 4.6 is to blame here - the problem >> is not there with 4.7. > > But we could not use qt 4.7 because of a font rendering bug? > > Will it be fixed or is this feature in a 'Done' component, as it is now > called by Nokia. It isn't a bug, it's a feature. Qt4.7 renders most fonts with kerning and ligatures as hinted by the font designers. LyX assumes fixed character width without kerning and so Qt draws the text different then LyX is positioning the cursor. That's why if LyX is linked with Qt4.7 the drawing is done character by character - the cursor positioning is ok then but the misspelled markers looks horrible because the dotted line is drawn for every single character too. Obviously much better then crashing... Stephan
