fudcon-room-5
LOGS
17:11:21 <boeckb> #startmeeting
17:11:21 <zodbot> Meeting started Sat Dec  5 17:11:21 2009 UTC.  The chair is boeckb. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:11:21 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
17:11:24 <loupgaroubl0nd> https://fedoraproject.org/wiki/FUDCon:How_To_Transcribe_a_Session
17:11:55 <OpenRebel> The Speaker has entered the Channel
17:12:22 <ianweller> boeckb: you doing transcribing?
17:12:33 <boeckb> #meetingtopic Effective Bug Reporting
17:12:49 <boeckb> ianweller: i was the one who asked who was doing it, so i voluteered by doing so
17:13:16 <ianweller> boeckb: thanks!
17:13:20 <adamw_> never volunteer for anything
17:13:36 <boeckb> #topic why file?
17:14:03 <boeckb> if not you, then who will?
17:14:12 <boeckb> helps contributing back to the community
17:14:49 <boeckb> #topic Where to report bugs?
17:15:11 <boeckb> Reporting to http://bugzilla.redhat.com to make packagers aware of issues
17:15:48 <boeckb> Reporting to upstream bug trackers http://bugs.kde.org and http://bugzilla.gnome.org also
17:16:14 <boeckb> #topic what to do before reporting
17:16:22 <boeckb> 1) search for duplicates
17:17:16 <boeckb> both in red hat bz then upstream
17:17:28 <boeckb> if it is upstream, file in red hat bz and link the bugs
17:17:53 <boeckb> #topic what to report against
17:17:55 <boeckb> example:
17:17:56 <boeckb> kate
17:18:06 <boeckb> is not a package in and of itself
17:18:21 <boeckb> yum provides *bin/kate
17:18:32 <boeckb> rpm -qf /usr/bin/kate
17:18:46 <boeckb> adam: rpm -qf `which kate`
17:18:57 <boeckb> adam: dont let it stop you from filing
17:18:59 <mchua> question from remotee: how do you remember the rpm -qf command? Is there a way to look up these things if we forget them after the session?
17:19:14 <boeckb> wrong product is better than no bug
17:19:43 <boeckb> adam: some people write them down
17:20:04 <OpenRebel> Answer: Sticky notes, man pages are help, what ever works best for you
17:20:21 <mchua> Is there a "handy commands for bug reporters" cheat sheet somewhere?
17:20:32 <boeckb> idea: put onto BugZappers wiki
17:20:38 <boeckb> (Steven)
17:20:53 <OpenRebel> mchua, There should be one after this talk
17:21:02 <mchua> Sweet :)
17:21:16 <boeckb> #topic determining release
17:21:37 <boeckb> yum list kdesdk
17:21:49 <boeckb> yum info kdesdk
17:22:01 <boeckb> both say which repository a package came from
17:22:11 <rescrv> look for the 'From repo' line in info
17:22:15 <adamw_> https://fedoraproject.org/wiki/BugZappers
17:22:19 <adamw_> (bugzappers wiki)
17:22:33 <boeckb> #topic What species is that?
17:22:34 <boeckb> keywords:
17:22:45 <boeckb> RFE (Request For Enhancement)
17:22:53 <adamw_> https://fedoraproject.org/wiki/BugsAndFeatureRequests <--- official page on how to file bugs
17:22:57 <boeckb> feature requests, package upgrades
17:23:08 <boeckb> Regression
17:23:08 <adamw_> we can put useful commands and so on into that, or as a sub-page of it
17:23:20 <boeckb> something that worked before but no longer works
17:23:50 <boeckb> Cosmetic bugs are minor errors (typos and the like
17:23:52 <boeckb> )
17:24:06 <boeckb> Other bug types
17:24:13 <boeckb> annoying bugs
17:24:18 <boeckb> usually have wrorkarounds
17:24:21 <boeckb> crashes
17:24:41 <boeckb> where the application crashes, usually/hopefully providing a backtrace
17:24:48 <boeckb> #topic extra information
17:24:56 <boeckb> what were you doing?
17:25:06 <boeckb> how easy is it to reproduce?
17:25:19 <boeckb> what happened (be as precise as possible)
17:25:31 <boeckb> what *should* have happened
17:25:50 <rescrv> what you expected to happen (in case you misunderstand developer's purpose)
17:26:14 <boeckb> if the devs cannot reprodce the issues, it is hard to fix
17:26:28 <boeckb> if it is graphics issues
17:26:33 <boeckb> include chipset and driver
17:26:56 <boeckb> extras:
17:27:00 <boeckb> screenshots
17:27:04 <boeckb> videos
17:27:07 <boeckb> command line output
17:27:20 <boeckb> stack trace (abrt can help)
17:27:35 <rescrv> start gui apps from cmdline and paste output
17:27:58 <comphappy> http://fedoraproject.org/wiki/StackTraces
17:27:58 <boeckb> adam: attach text as text, not compressed
17:28:06 <boeckb> #topic upstreaming bugs
17:28:20 <boeckb> getting the bugs known to those who will do the fix
17:28:39 <boeckb> maintainers help provide information, but the reporter knows the bug best
17:29:34 <boeckb> if you find a match, add a comment
17:29:57 <boeckb> new bug if it isnt there
17:30:07 <boeckb> link the red hat bug to the upstream bug
17:30:22 <boeckb> adam: UPSTREAM keyword
17:31:18 <boeckb> #topic stack traces
17:31:38 <boeckb> programs have extra information in a -debuginfo package
17:32:09 <boeckb> this offers extra information about what the program was doing before it crashed
17:32:18 <boeckb> debuginfo-install <package>
17:32:32 <boeckb> installs all relevant debuginfo packages
17:32:50 <boeckb> #topic using gdb to get a stack trace
17:33:00 <boeckb> gdb <name of program>
17:33:14 <boeckb> at the prompt:
17:33:17 <boeckb> run
17:33:31 <boeckb> add command arguments to run
17:33:44 <boeckb> thread apply all to get the backtraces of all threads
17:34:04 <boeckb> when it crashes, the "bt" command prints the backtrace
17:34:27 <boeckb> #topic watch your report
17:34:32 <boeckb> other's comments
17:34:38 <boeckb> needinfo flag
17:34:49 <boeckb> developers requesting information from users
17:35:10 <boeckb> usually involves logs, command output, or backtraces
17:35:29 <boeckb> after 30 of no information offered, bug is usually closed as INSUFFICIENTDATA
17:35:47 <rescrv> 30 days that is
17:35:49 <boeckb> reopen the bug if this happens and you have the information
17:35:53 <boeckb> resolution
17:36:07 <boeckb> when the fix is made, the bug may be set to MODIFIED
17:36:36 <boeckb> and testing may be needed before the fix can make it to the repositores
17:37:10 <boeckb> attendee: what about abrt?
17:37:18 <boeckb> (pascal)
17:37:32 <boeckb> daemon that watches for application crashes
17:37:43 <boeckb> puts a caution icon in the tasktray
17:37:54 <rescrv> works for python/c++
17:38:09 <boeckb> files bugs (still needs an account and extra information)
17:38:20 <boeckb> can also install debugiinfo packages automatically
17:38:43 <boeckb> pascal: kerneloops go to somewhere else?
17:39:04 <boeckb> yes, they are anonymous traces for the kernel
17:39:20 <boeckb> adam: kernel developers look for trends
17:39:53 <boeckb> lots of a crash means more chance of fixes
17:40:36 <rescrv> sometimes USB devices can cause kerneloops
17:40:48 <rescrv> abrt will sometimes grab insignificant messages
17:40:58 <rescrv> still a work in progress and open to suggestions
17:41:24 <rescrv> to get involved, join the bugzppers group @ 1530 UTC on Tuesdays
17:41:42 <boeckb> (1500 is official, KDE-SIG runs late usually)
17:42:10 <boeckb> sorry about that (from KDE SIG) ;)
17:42:34 <boeckb> protip: dont get offended by commnts on bug reports
17:43:05 <boeckb> adam: be nice. even if you lost your thesis, the developer is more amenable to nicer bug reports
17:44:05 <boeckb> be aware that it is fedora you are running
17:44:36 <boeckb> adam: people are afraid of having their bug being ignored
17:44:45 <boeckb> adam: 60% chance if you file
17:44:50 <boeckb> 0% chance if you dont
17:45:40 <boeckb> faster responses make maintainers more responsive
17:46:11 <boeckb> (brendan)
17:46:16 <comphappy> Brennan Ashton
17:46:49 <boeckb> anaconda needs testers and bugs filed
17:47:16 <boeckb> doesnt get much testing due to its nature
17:47:22 <boeckb> of being an installer
17:49:28 <boeckb> #endmeeting