Greg Ercolano wrote: > MacArthur, Ian (SELEX GALILEO, UK) wrote: >>> Today Grigoriy (the OP of STR 2185) asked in fltk.bugs: >>> > So when this fix will go to svn? > >>> Should we remove the comment and see what happens, as I proposed >>> above two weeks ago? > >> I think we could take the chance and see what happens. > > http://www.fltk.org/str.php?L2185 > Ahh yes, I sleuthed that one, so I feel compelled to respond.. > > I tentatively agree, with the stipulation that we probably need to > advertise (or fix if possible) the problem where Grygoriy determined > he *could* replicate fltk apps *crashing the entire xserver* with a > certain version of xorg (7.2). > > See his 09:13 Apr 02 2009 entry with the backtrace in STR#2185.
IMHO Grygoriy's comments showed that his problem was unrelated to the XCloseIM calls (commented or not). He wrote: "Then I downgraded to xorg-x11-7.2 and now I'm experiencing crash of xserver on xorg-x11-7.2 regardless of presence of XCloseIM calls. I tried to comment all XCloseIM calls in all fltk source files. But xserver still crashes. This happens when I'm trying to enter some text into Fl_Input. If I compile fltk without xft everything works fine." > With that info, maybe no one felt too excited to uncomment, as it > probably means fltk apps will crash the entire xserver on linux distros > that have that 7.2 xorg release (or whatever /range/ of xorg releases > has this problem, 7.2 just being one) > > All we really know is that 7.2 crashes, and 7.4 does not. > > Probably our best X windows font guy should look at this to see > if it can be prevented, since it can take down the whole window manager: Interesting side note: Today I experimented with FLTK 1.3 on Ubuntu 8.10. Although FLTK itself worked without a crash (and I did also uncomment the XCloseIM call and reproduced the foo2.cxx behavior in both cases), I had a problem with my own FLTK application. It crashed at a point where I would never expect it, but the strange thing was: I couldn't really debug it, because the crash didn't happen, when compiled without optimization (-g -O0), but happened (with or w/o -g), when compiled with -O1 or higher. My suggestion is: uncomment the XCloseIM call now (better now than after we went public), and see what happens, but do also investigate the other issues. I'll debug my problem further, but I don't know when I'll find the time. Probably not before the weekend :-( Albrecht _______________________________________________ fltk-dev mailing list [email protected] http://lists.easysw.com/mailman/listinfo/fltk-dev
