[Libreoffice-qa] LibreOffice QA Call Metting Minutes 2012-09-07
Petr Mladek
pmladek at suse.cz
Fri Sep 7 09:10:37 PDT 2012
Hi,
Participants: Florian, Petr
We had great chat but really missed the others. Please, try to join next
time or say what is the blocker to join.
pending action items:
- write update scenario testcase in Litmus/MozTrap (Kendy)
- Invite active bugwranglers to next call/QA list, CC Rainer (Bjoern)
- merge 3.5 and 3.6 in one big bibisect repo (Bjoern)
- recheck and tweak bibisect details (Bjoern)
- Ping cloph if we can make that switchable to say "Bug" instead of
"EasyHack" when explcitly requested (Bjoern)
- new Bug Hunting Sessions (Cor)
* noting new; The responsible people were missing :-(
updated schedule
- http://wiki.documentfoundation.org/ReleasePlan
- two weeks between RCs
- 3.7.0 and 3.7.1 collision
* [Petr] We added one week between RCs; the only exception is 3.7.0-rc2-rc3
and 3.7.1-rc1-rc2 because they conflicted, see also
http://lists.freedesktop.org/archives/libreoffice-qa/2012-September/002514.html
* [Florian] it looks like the best solution
* [Petr] we are still improving the schedule; the current one has the drawback
that there is a short time between the 3.X.Y release and the deadline for
3.X.Y+1-rc1; it would work if we do not add regressions; the bugfixing process
is a kind of pipeline anyway; though, we might want to have bugfix release every
5 weeks instead of every 4 weeks; let's see how the current schedule work
in practice
structured manual testing (Yifan/Petr):
- localization
* Yi Fan proposed 3 temporary solutions; Sophie provided feedback from French
community; so we selected one solution, see
http://lists.freedesktop.org/archives/libreoffice-qa/2012-September/002485.html
- 3.6.3 testing
* Sophie would like to use Moztrap for 3.6.3 testing
* Yi Fan is moving testcases from Litmus and fixing usability
problems to make it ready
* any help or feedback is welcome
- presentation on the LO conference
* Petr will do presentation/workshop about Moztrap on LO conference
because Yi Fan fill not be there
bug wrangling (Rainer):
* Florian showed some bugzilla statistic produced by Rainer:
* [Florian] Big number of bugs reported against 3.7.0.4;
many must be older bugs => should we ask users to try
older versions and set the oldest affected?
* [Florian] Also many bugs has the version "Unspecified";
should we ask for the affected one?
* [Petr] IMHO, the most important information is how to
reproduce the bug; the version is nice to have; it helps
when the bug is not reproducible with some other build;
also it helps to locate the problem if it is a regression
=> we should ask for the version and set it; also we should
try older versions, attach backtrace, test documents, do bibisect
but all this is only nice to have; if the bug is well reproducible,
we could pass it to the development immediately and add the helpful
details later
Well, other people might see it different; feel free to reopen this
question on the next call
(to be continued)
* Florian also got idea to arrange "bug hunting hours"
* it might work better than days
* it should be during week when fultime developers
are around
=> let's discuss this on the next meeting with more people;
we need to find a good time, ...
(to be continued)
- bugzilla contract improvements (Rainer)
* We missed Rainer
- last 5 hardhacks retrospedctive:
this table is taken from ESC call meeting minutes
--- cut ---
ID OS Component - Summary
--------------------------------------------------------------------
38913 Windows Libreoffice - CRASH when loading Danish dictionary at
startup citing msvcr90.dll after upgrade
+ potentially jvm dll at fault, use a workaround to
avoid fatal crash when called API called with an
illegal argument.
+ fixed - thanks to Stephan, custom handler works
around crash.
32948 Linux Libreoffice/Database(?) - Address Book Data Source
Wizard fails with message "No SDBC driver was found"
+ bug-log mixes several different issues, very unclear.
+ please open separate, new, clean issues & re-report.
+ one is split to bug#54450 Evo/Addressbook doesn't work
+ David to kindly look into it.
AI: + enable evolution connector in distro-config for Linux (Lionel)
34548 All Presentation - EDITING: CRASH in action after Undo
+ valgrind trace preferred.
36681 All Writer - EDITING: after insert/removal of a picture (or
alike) view scrolls to begin/end of document
+ nasty cases where picture anchored to the page, and
inserting paragraphs afterwards, screen showed the 1st
page - due to stale layout info.
+ fixed by Cedric.
33302 Mac Libreoffice - FILEOPEN/EDITING RTL text: parentheses and
brackets "(...) [...]" inverted to ")...( ]...["
with some fonts
+ Mac volunteers appreciated.
--- cut ---
=> 2 fixed; 1 put back to QA; 2 not really touched
* [Petr] nice start; development might want to assign the bugs during
ESC meeting to make sure that we move forward with the two weeks
- find another 5 HardHacks
* [Petr] let's keep the two non-solved:
* https://bugs.freedesktop.org/show_bug.cgi?id=34548:
EDITING: CRASH in action after Undo
* https://bugs.freedesktop.org/show_bug.cgi?id=33302:
FILEOPEN/EDITING RTL text: parentheses and brackets "(...)
[...]" inverted to ")...( ]...[" with some fonts
* after long search we nominated these 3:
* https://bugs.freedesktop.org/show_bug.cgi?id=44664:
cups landscape woes with pdf: printer truncates Landscape
orientation pages to Portrait dimensions
* https://bugs.freedesktop.org/show_bug.cgi?id=35680:
FORMATTING: Gradient rendered wrongly in presentation mode when grouped
* https://bugs.freedesktop.org/show_bug.cgi?id=39007:
Persistent "Restore Windows" dialog won't disappear, disables key functionality
They all are MABs, long standing, have many people in CC, many duplicates, ...
* Please, join the call next time and help us to find the right hardhacks :-)
- bug prioritization, can we start for real (continue)?
http://nabble.documentfoundation.org/Libreoffice-qa-Bugs-prioritization-missing-pieces-td4002450.html
* [Florian] Only experienced people should be able to modify these fields;
normal people do not see the wide picture and set high severity and priority
for their pet bugs
* [Petr] Yes, this might be a problem. Well, we should correct the severity/priority during
the bug triage; Few users will not agree and increase it; We could try to set it back
once again and then ignore them; if only 1% of bugs has wrongly set priorities, it still
would be great state
* [Petr] Important is to decide on the rules; All bug triagers should be on
the same wave, so they do not fight against each other
(to be continued)
community building/communication:
- new time for QA call retrospective
* [Florian] the new time is fine now; I am not sure how it will be
once the next semester starts; let's see
- Joel's new page describing the BugTriage process
- https://wiki.documentfoundation.org/BugTriage_InProgress
* we missed Joel
- Nino picked my mail and put excepts up as a QA mission statement
- http://nabble.documentfoundation.org/Libreoffice-qa-QA-Mission-Statement-was-Re-Fwd-Re-Closing-NEEDINFO-bugs-td4002426.html
- http://wiki.documentfoundation.org/QA#Bug_Triage.5B1.5D
- what is the message for a newcomer?
- is there a more clear and encouraging alternative?
* [Petr] Especially the sentence "Given that 87 of the assigned
bugs have not seen a change in more than 180 days" sounds depressing.
It is true but it does not mean the developers did not fix bugs;
They just could not fix all the 5000 bugs within weeks.
The sentence might discourage new triagers that the work does
not make sense because developers does not care. This is not true.
Developers fox bugs. The triaging is really important. Also the
prioritization is important.
* We went completely out to time, so we did not come up with
the better formulation.
(to be continued)
Best Regards,
Petr
PS: Florian, I am sorry if I missed some of your great ideas. My English
is not 100%. Also I feel a bit tired.
PPS: I hope to hear more people on the call next time. It is really
great opportunity to exchange opinion and decide on some things faster.
More information about the Libreoffice-qa
mailing list