GNOME Bugzilla – Bug 118641
gnumeric crashes trying to start it.
Last modified: 2004-12-22 21:47:04 UTC
Distribution: Red Hat Linux release 9 (Shrike) Package: Gnumeric Severity: normal Version: GNOME2.3.4 1.1.19 Gnome-Distributor: GARNOME Synopsis: gnumeric crashes trying to start it. Bugzilla-Product: Gnumeric Bugzilla-Component: Analytics Bugzilla-Version: 1.1.19 BugBuddy-GnomeVersion: 2.0 (2.3.3.1) Description: Description of the crash: Clicking the gnumeric icon or starting it from the terminal results in a crash. I have had it running after installing it using garnome, but it no longer works. Not sure what changed. Debugging Information: Backtrace was generated from '/home/rodd/gnome-2.3.3/bin/gnumeric' [New Thread 1091392192 (LWP 11861)] 0xffffe002 in ?? ()
+ Trace 39143
Thread 1 (Thread 1091392192 (LWP 11861))
------- Bug moved to this database by unknown@bugzilla.gnome.org 2003-07-30 03:23 ------- Unknown version 1.1.19 in product Gnumeric. Setting version to the default, "unspecified". Reassigning to the default owner of the component, jody@gnome.org.
Gnome 2.3.4? Gnumeric 1.1.x is written for Gnome 2.0 or 2.2.
Lets send this over to linc to see if that trace rings any bells. This is well before gnumeric gets control of things. Something is pissing linc off. Are you loading some gtk modules ? you can try wiping your ~/.gconf/apps/gnumeric dir to see if that makes gconf happy, or deleteing /tmp/orbit-<username> to see if there is something dangling in there. andreas : gnome-2.3.4 is fine, we work againt cvs head too.
For what it's worth, this appears to be a unique stack trace. Setting severity->critical (crasher), adding bugsquad and STACKTRACE (due to the debugging symbols) keywords, and selecting "Reassign bug to owner of selected component" (/me nudges Jody to remind him to do that as well when reassigning).
Okay, I tried deleting ~/.gconf/apps/gnumeric and no change, but when I deleted /tmp/orbit-<username> this fixed it. Unfortunately, I didn't stop and think maybe to move it so that I could supply it to you to see what was wrong.
Very odd; it is _possible_ but unlikely that there was an insufficiently random number and we got a socket collision in /tmp/orbit-$USER, but we have code to catch and avoid this. I've really no idea how a valid, listening socket can get in such a state; we do our listen before setting O_NONBLOCK; really no idea.
a duplicate it seems: *** This bug has been marked as a duplicate of 113020 ***