[Libreoffice-qa] minutes of ESC call ...
Michael Meeks
michael.meeks at collabora.com
Mon Oct 19 02:17:16 PDT 2015
* Present:
+ Stephan, Jan-Marek, Eike, Kendy, Miklos, Michael M, Andras, Christian,
Norbert, Robinson, Caolan, Bubli, Bjoern, Muthu
* Completed Action Items:
* Pending Action Items:
+ please check if you're listed in the mentoring page (All)
+ https://wiki.documentfoundation.org/Mentors
[ still needs some action from Kohei, Cedric, Thorsten etc. ]
+ UserAgent - drop bundled-languages (Michael)
+ UserAgent - produce a patch for review (Michael)
+ ask students to write their own wrap-ups to aggregate (Cedric)
+ write a nice blog entry about 1000's bibisect - including flowers (Robinson)
[ updating a graph, will post today ].
+ reach-out to Beta Co-working guy to book post-FOSDEM hackfest (Bjoern)
* Release Engineering update (Cloph)
+ 4.4.6 - RC2 status
+ tagged yesterday, builds are up-loaded for Win, Linux & Mac
+ can be announced for early-testing
+ available for all tomorrow.
+ 5.1.0 Alpha - this week ?
+ done either today or tomorrow.
+ 5.0.3 - RC2 status
+ due next week - tag planned towards end of week,
Thur evening
+ Android & iOS Remote
+ promoted Alpha -> Production for the Google app-store.
+ migrated build-system to gradle
+ supports debugging with Android Studio & lldb
+ if not giving the same results as ndk-gdb - try the other
+ they should both work (see android/README for details)
* Killing MediaWiki-Publisher (Andras)
https://gerrit.libreoffice.org/#/c/19167/ => seems to be solved
+ not killed after all, just removed deps. to old & buggy bits.
=> modernized.
* Killing LibreOffice gconf integration (Bjoern)
+ gconf is obsoleted by gsetting
+ on most distros the settings in gconf will be outdated/unused
+ keeping this around as will likely just cause confusion as people will find subtile different settings without knowing why/from where
+ https://bugs.documentfoundation.org/show_bug.cgi?id=41678
+ used to be used for lockdown and some prop. tools integrated (Michael)
+ RH not using it (Caolan)
+ Stephan working on a dconf version instead.
* UX Update (Kendy)
+ NotebookBar
+ https://wiki.documentfoundation.org/Development/NotebookBar
+ merged the prototype to master from the branch, guarded
by an env. var (export LO_USE_NOTEBOOKBAR=1)
+ some extensions of the .UI XML to accomodate this nicely.
+ Samuel working on Kendy's prototype here
+ Breeze / HiDPI theme work ongoing.
+ Tomaz merged the darkening icon-theme algorithm, design guys happy.
+ Table Styles
+ a new design session on Friday
+ last time discussed editing
+ now necessary to approach applying them in the sidebar etc.
+ keep toolbars or move to the sidebar ?
* Documentation (Kendy)
+ again some new patches from Jay for help-authoring ext'n.
+ will release a new version tomorrow.
+ #libreoffice-design IRC a great place to hang out to implement help.
* Crashtest update (Caolán)
+ 1 (now fixed) import failures, 0 export failures
+ doing another run again.
+ if still zero - need to try asan build.
+ coverity upgraded to 7.7.0, now c++ *and* java
+ project broke since marked C++ only.
+ 657 warnings!
+ mostly java or c++ rtti change related
+ what warnings do we get from RTTI ? (Michael)
+ dynamic_cast<> can return NULL (Caolan)
+ dynamic on const, and then static_casts etc.
+ helps find places where we can clean casts up.
* File format fuzzing to drive UX testing (Caolan)
+ video: http://caolanm.blogspot.ie/2015/10/fuzzing-libreoffice-input-events-with.html
+ file format driving UI input
+ got the server load piece done;
getting iterations from under 1x per sec. to 4x per sec.
+ file-formats went to several 1000 per sec.
+ events very far removed from consumption, so hard to
see the chain between consumption and insertion.
* Hackfests (Bjoern)
+ Hamburg, Germany: Oct 23rd -> 25th (Bubli)
+ https://wiki.documentfoundation.org/Hackfest/Hamburg2015
+ would be great to see everyone there.
+ Friday will celebrate the 5th birthday again with
misc. parties / politicians joining in the fun.
+ FOSDEM / Beta Co-working
+ do it pre-FOSDEM ? (Eike)
+ does it work to arrive 2x days earlier ?
+ slide-writing instead ? (Michael)
+ not much communication happening anyway (Eike)
+ lots of tired people.
+ arn't really supposed to talk a lot @ co-working (Miklos)
+ could we book a separate room for us ? (Bubli)
+ afterwards - could invite people to join in ? (Bjoern)
+ perhaps try it beforehand this time.
+ next venues / suggestions
+ Belgium another candidate ...
+ perhaps again at Las Palmas (Thorsten)
* l10n Update (Sophie)
+ did register #libreoffice-NLP channel - join it ...
+ rather an unfortunate name (Bubli)
+ expecting 'l10n' rather than nlp.
+ asked on the l10n list for feedback, people preferred NLP (Cloph)
AI: + could setup a notification message for -l10n -> NLP (Cloph)
* Developer Certification (Stephan/Bjoern/Kendy/Thorsten)
+ now we have 5 names
+ some new suggestions to come from Bubli.
+ reaching out to poke the happy recipients etc.
+ another round planned in a month or two.
* Jenkins update (Norbert)
from:Thu Oct 8 13:53:52 2015
master linux rel jobs:208 ok:196 ko:12 fail ratio:5.76% break:4 broken duration:3.92%
master linux dbg jobs:122 ok:114 ko:8 fail ratio:6.55% break:5 broken duration:3.66%
master mac rel jobs:185 ok:173 ko:12 fail ratio:6.48% break:6 broken duration:2.83%
master mac dbg jobs:194 ok:176 ko:18 fail ratio:9.27% break:11 broken duration:7.92%
master win rel jobs:93 ok:81 ko:12 fail ratio:12.9% break:8 broken duration:14.9%
master win dbg jobs:160 ok:126 ko:34 fail ratio:21.24% break:23 broken duration:15.92%
master win64 dbg jobs:136 ok:115 ko:21 fail ratio:15.44% break:15 broken duration:12.97%
lo-4.4 mac jobs:8 ok:7 ko:1 fail ratio:12.49% break:1 broken duration:23.45%
gerrit win rel jobs:212 ok:124 ko:87 fail ratio:41.03%
gerrit lin rel jobs:212 ok:182 ko:30 fail ratio:14.15%
gerrit mac rel jobs:212 ok:166 ko:46 fail ratio:21.69%
gerrit all jobs:212 ok:113 ko:98 fail ratio:46.22%
+ an average week.
+ do we have any windows machines with symbol-server working ? (Caolan)
+ not thought that there are any (Cloph)
+ '39' tbox - not working, unit tests not working (Kendy)
AI: + will try to re-start that now (Kendy)
+ can we get a README for setting that up ? (Norbert)
+ nothing that difficult, either documented or out-of-box (Kendy)
+ ideally should be running on TDF infra (Norbert, Cloph)
+ is part of default tinderbox scripts (Cloph, Norbert, Kendy)
* Easy Hacks:
+ new easy-hacks thanks to:
+ Sophie, Michael
+ http://bit.ly/1KR4HWX
+ create your own by setting whiteboard to eg.
"easyHack difficultyBeginner skillCpp topicCleanup"
+ create new easy hacks from these proposed ones
(or remove the whiteboard keyword "needsDevEval" if not feasible):
https://bugs.documentfoundation.org/buglist.cgi?status_whiteboard=needsDevEval
* include ordering (Miklos)
+ some people have strong opinions on canonical include ordering
+ when its a mess - adding new ones at the end a bit annoying.
+ can someone (who cares about this) write up some suggestions ?
+ tool used to remove unused includes (?)
+ sal/config.h should be the 1st include (Michael)
+ new contributors need to know this stuff (Miklos)
+ there is documentation somewhere - talking about sal/config (Norbert)
+ a rule almost never followed; but technically good.
+ after that - avoid a mixed up zoo.
+ some sort of rough ordering; system, external modules, local
+ easier to read & maintain.
+ another rule: (Bjoern)
+ after sal/config.h - include header corresponding to your file
+ as the 1st include; so the header is self-contained & complete.
+ even this rule questionable (Stephan)
+ lets not bike-shed.
+ problematic to cleanup by re-ordering with no real merit (Stephan)
+ conflicting cleanups that revert each other's patches (Miklos)
+ some is cosmetic, but not all can be resolved technically (Norbert)
+ cosmetic -> multiple ways to do it; just one way.
+ what the old OOo code conventions say about it: https://wiki.openoffice.org/wiki/Cpp_Coding_Standards/HEADERS
+ "include what you use" triggered this (Miklos)
+ if so - reducing pointless churn a goal (Michael)
+ also improving the include hierarchy
=> leave it for now.
* QA update (Robinson)
+ UNCONFIRMED at 483
+ lower.
+ needAdvice at 15
+ thanks for the VclPtr bisection work from Bjoern (Michael)
+ Most Pressing Bugs:
http://ur1.ca/nkp0t
+ tdf#94063 - EDITING References break when putting text behind them
+ tdf#91789 - UI: Crash on exit after format header
+ update from Bjoern (fixed?)
+ tdf#94689 - Crash on new file, close file, open file
+ Another windows crash
+ tdf#94225 - Writer crashes on undo
* QA stats
+ https://bugs.documentfoundation.org/page.cgi?id=weekly-bug-summary.html
+206 -279 (-73 overall)
many thanks to the top bug squashers:
QA Administrators 93
Beluga 20
Julien Nabet 17
m.a.riosv 15
Maxim Monastirsky 14
Joel Madero 13
Markus Mohrhard 10
V Stuart Foote 9
Dennis Roczek 6
raal 5
Adolfo Jayme 4
Miklos Vajna 4
Cor Nouws 4
Jacques Guilleron 4
Winfried Donkers 3
* Raw data for new metrics:
+ issues marked as 'highest' priority.
+ considered equivalent to MAB.
5.0: 9/35 - 25%
4.5: 3/12 - 25%
4.4: 11/70 - 15%
4.3: 7/67 - 10%
4.2: 12/132 - 9%
4.1: 4/79 - 5%
4.0: 12/81 - 14%
3.6: 5/41 - 12%
3.5: 5/63 - 7%
3.4: 6/58 - 10%
3.3: 6/34 - 17%
OO: 16/28 - 57%
http://tdf.io/mab
* Open 5.0 MAB
+ 12/47 [new metric] 1/12
* Open 4.4 MAB
+ 68/70 [ new metric ] 8/59 10/60
* bug stats (Robinson)
+ previously - hand-chosen; tagged on earliest live series.
+ would get moved forward when we retired the series.
+ here, we have them sitting as far back as we can trace.
+ can we add the 4.5 and 5.0 together ? (Michael)
+ yes (Robinson)
+ can we have a 5.1 number ? (Michael)
+ yes (Robinson)
+ can we have a sum of all still open bugs for
all versions vs. closed for this version (Michael)
+ by release series? sure (Robinson)
+ is something still open since 3.3 truly a MAB ? (Michael)
+ worth looking at again ?
AI: + discard MAB series (Michael)
+ closed bugs are pretty random (Bjoern)
+ new set of graphs from next week.
* Bibisected bugs open: whiteboard 'bibisected'
+ 324/1064 322/1047 324/1039 325/1029 330/1028 344/1028 345/1020 338/1006
+ http://tdf.io/bibibugs
* all bugs tagged with 'regression'
+ stats are for two weeks ...
+ 683(+6) bugs open of 4327(+84) total 48(+0) high prio.
* ~Component count net * high severity regressions
Calc - 19 (+1)
Writer - 16 (+0)
LibreOffice - 7 (-1)
Impress - 3 (+0)
UI - 1 (+0)
filters and storage - 1 (+0)
Base - 1 (+0)
+ http://bit.ly/1HWHb3E
* ~Component count net * all regressions
Writer - 233 (+5)
Calc - 143 (+5)
Crashes - 52 (-1)
Impress - 49 (+0)
UI - 45 (+1)
LibreOffice - 45 (+0)
Borders - 27 (+3)
graphics stack - 26 (-1)
Base - 26 (-1)
Draw - 21 (-1)
filters and storage - 15 (+2)
Chart - 12 (+0)
Printing and PDF export- 11 (+0)
BASIC - 10 (+1)
Formula Editor - 5 (+0)
Linguistic - 3 (+0)
framework - 3 (+0)
Extensions - 2 (+0)
sdk - 1 (+1)
+ http://bit.ly/1BUdI8i
--
michael.meeks at collabora.com <><, Pseudo Engineer, itinerant idiot
More information about the Libreoffice-qa
mailing list