GNOME Bugzilla – Bug 383349
reject empty stacktraces in bug-buddy already
Last modified: 2006-12-14 18:14:01 UTC
quoting luis from the comments to http://blogs.gnome.org/view/aklapper/2006/12/06/0 , filing this as a bug report after talking to guenther: "why aren't those completely useless stacktraces being parsed and rejected by bug-buddy? Seems like a trace as empty as the one you linked to[1] should never leave the PC. Perhaps, best case, you say 'unfortunately, there is no useful automated information available. If you can give detailed information on what caused the crash, please do: _______________ [submit] otherwise, please [restart] Or something like that." [1] http://bugzilla.gnome.org/show_bug.cgi?id=381577 the auto-rejecter cannot parse on this, and it's already too late because the bug has been committed already. this has to be rejected at the bug-buddy stage already, by checking whether there is any content right after the "Backtrace was generated from '/usr/bin/eog'" or the "Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1"." line. if there is not, we should tell the user, and that he should install debug packages by using the package manager of his distribution. however, we must make sure that we do not reject enhancement requests, but as i have understood, we now use bug-buddy for crashers only (please correct me if i'm wrong).
Thanks a lot for the report. I'm doing a trick searching for "#1", "#2" and "#3" gdb marks, if they are not present the backtrace is useless and we are pointing the user to install debug packages. This is a temporal solution of course until we get the useful traces without debug symbols thanks to airbag/debug server. 2006-12-13 Fernando Herrera <fherrera@onirica.com> * src/bug-buddy.c: (gdb_insert_text), (create_debuginfo_link), (useless_finished), (gdb_finished): Don't submit reports with less than 3 steps on the backtrace. Fix bug#383349
thanks so much!, from a triager.
*** Bug 385931 has been marked as a duplicate of this bug. ***