On Tue, Oct 07, 2003 at 07:57:52PM -0400, Grzegorz B. Prokopski wrote: > Apparently yes. I've just upgraded my system and tried again from GNOME > menu. Same result. Here's the relevant part of this log file. > > /usr/lib/openoffice/program/libvos3gcc3.so(_ZN3vos14OSignalHandler5raiseElPv+0x26)[0x40b443c0] > /usr/lib/openoffice/program/libtl645li.so[0x409b9a10] > /usr/lib/openoffice/program/libtl645li.so(_ZN6ResMgr4InitERK6String+0x24)[0x409b9bee]
Looks like a crash in RegMgr::::Init(). Grzegorz, I've made a debug version of that shared library: http://people.debian.org/~halls/openoffice/1.1.0/debug/libtl645li.so Please could you replace the library in /usr/lib/openoffice/program with this one, and reproduce the crash. At the moment when the dialog is displayed within OOo, attach to the process using gdb --pid=<pid-of-soffice.bin> /usr/lib/openoffice/program/soffice.bin and get a backtrace (thread apply all bt) and post it? That should let me pinpoint exactly which line is crashing. > Just started oowriter from multi-gnome-terminal and it worked. > 6 is SIGABRT so smells like if some assert() was failing but it's > only my speculation. asserts aren't compiled in the packages, but the code in ResMgr does tend to blow up if something goes wrong, and the abort will be caused by OOo catching a SEGV, 'handling' the crash and then calling abort() from within the handler. Chris
