minutes of ESC call ...
Michael Meeks
michael.meeks at collabora.com
Thu Apr 21 16:29:40 UTC 2016
* Present:
+ jan i, Lionel, Kendy, Michael M, Stephan, Armin, Miklos, Andras, Bubli
Caolan, Christian, Sophie, Bjoern, Muthu, Norbert, Michael S
* Completed Action Items:
* Pending Action Items:
+ poke again at the lcov problem on tb31 ? (Bjoern)
+ poke at hack-fest candidate list (Bjoern)
+ investigate clobberring glew instead of lots of #ifdefs: (Marco, Oliver)
[ Working on getting this up-stream ]
+ tweak UI and get LiveConnect API key / build case for board (Christian)
+ ask QA to re-test http://tdf.io/mmregressions (Robinson)
[ testing underway ]
+ attempt to re-build a recent gstreamer 1.0 / core spec file
on our CentOS6 base (tdf#94508) (Christian)
+ script XML / l10n string translation changes (Christian)
[ have a partial script - but - waiting for pootle migration to
finish -> next week. ]
* Release Engineering update (Christian)
+ 5.0.6 RC2 - due for tagging next Tues -April 26th
+ still patches needing tripple review
https://gerrit.libreoffice.org/#/q/status:open+project:core+branch:libreoffice-5-0
→ don't forget to also cherry-pick for libreoffice-5-0-6 branch
+ 5.1.3 RC1
+ on the mirrors, can be announced later this week
+ Linux build ran out of space had to be cleaned up
+ 5.2 Alpha 1
+ tagged earlier today, builds are done
+ ready for bug-hunting session *tomorrow*
+ a milestone for early testing.
+ Android & iOS Remote
+ expect Viewer update after Alpha a tagging
+ Aleksandar working on theming for Android remote.
* ESC composition
+ Slightly nominal position:
+ https://wiki.documentfoundation.org/Development/ESC
+ https://wiki.documentfoundation.org/TDF/Engineering_Steering_Committee
AI: + build a proposal & reach-out for up/down next time (Michael)
+ mail me privately / ESC list if you have concerns.
* Documentation (Olivier)
+ Working on update for Chapt. 4 of book Getting Started - Writer
+ worked on XSLT transformations for online help.
+ http://vm173.documentfoundation.org
+ goal for publishing it ? (Michael)
+ need help from a skilled HTML developer
AI: + good to have a plan of when this is published (Olivier, Kendy)
+ does it go live with 5.2 ?
+ tender for editing XML on the side (Olivier)
+ would be really nice.
+ lots of budget commitments currently (Kendy)
* l10n (Sophie)
+ all pending migration
+ a number of fixes, and a security update coming to pootle
* GSoC 2016 (Bubli)
+ Tomorrow results announced.
+ community bonding period until May, 22nd
+ still need back-up mentors for some tasks
* UX Update (Kendy/Bubli)
+ Sidebar feedback write-up / blog-post ~finished
+ how should we handle ux-advice in bugzilla ?
+ currently a separate component
+ doesn't work so well; moving from Draw -> ux-advice etc.
+ reset from darw -> UX advice & then can't be filtered.
+ in BZ - can we bind a keyword to automatic adding of CC ?
+ would be lovely to have a keyword instead.
+ could we add a trigger for this somehow ?
+ it can be done (JanI)
+ seen it in a newer version.
AI: + look into it (JanI)
+ not terribly urgent, but good mess to fix (Kendy)
+ also for easyHacks (jani)
* Crashtest update (Caolan)
+ 0 import failures, 0 export failures, 0 coverity
+ excellent numbers
+ some random fluctuation suspected
+ will re-sync. with BZ - and up the numbers.
* Hackfests (Bjoern)
+ next hackfests:
+ Ankara, Turkey
+ April 30th - May 1st.
+ https://wiki.documentfoundation.org/Hackfest/Ankara2016
+ plan do an overview talk on Friday with Markus & any interest (Michael)
+ next weekend - should be fun !
+ Germany, Munich, LiMux; End of May 2016 (Jan-Marek)
+ Orga: https://wiki.debian.org/BSP/2016/05/de/Munich
+ 27-29 May (parallel to the Debian Sun Camp)
+ next venues / suggestions
+ https://www.froscon.de/en/home/Flosscon in Bonn, Germany
+ August - 2 weeks before LibreOffice conference ?
+ perhaps piggy-back on their rooms.
+ like FOSDEM at a university etc.
* LibreOffice Conference in Brno - CfP open (Sophie)
+ Kendy & Thorsten on program committee:
+ https://blog.documentfoundation.org/blog/2016/04/08/libreoffice-brno-conference-call-for-paper/
+ deadline is mid July.
* mentoring/easyhack update (janI)
+ contributor means a not-yet-committer.
+ statistics, numbers in () are +/- compared to last week
+ gerrit: committer open 64(-9) contributor open 59(-4)
updated merged abandoned updated merged abandoned
year: 64(-9) 6112(23) 545( 8) 59(-4) 1408(14) 460(-4)
month: 62( 2) 594(54) 36( 6) 58(-4) 164( 1) 85( 1)
week: 35( 0) 150(-8) 15( 8) 19(-6) 35( 0) 12(-7)
+ openhub: commits year 18239(-) developers to mail 48(-)
people with commits year 299(-) month 75(-) week 21(-)
+ easyHacks:
total 285(-20) open 243(-32) assigned 38(25)
needinfo 4(0) needsDevEval 433(4)
missing cleanup due to many comments 219(-10) control still valid 198(-1)
+ top 5 contributors (based on merged patches last month):
Jochen Nitschke 35 of total 44 merged patches
Rohan Kumar 15 of total 42 merged patches
Tamás Bunth 11 of total 19 merged patches
Muhammet Kara 10 of total 11 merged patches
Douglas Mencken 9 of total 36 merged patches
+ new easyHacks:
https://bugs.documentfoundation.org/show_bug.cgi?id=99341 mentor:fito -> "Simplify Document Recovery dialog by removing unnecessary header widget"
https://bugs.documentfoundation.org/show_bug.cgi?id=99311 mentor:bjoern.michaelsen -> "Kill Pagein for good"
+ create your own by updating Keywords to eg. "easyHack, difficultyBeginner, skillCpp, topicCleanup"
+ please remember, a codepointer is mandatory, and typically a look from an experienced developer
+ create new easy hacks from these ones or remove "needsDevEval" if not feasible):
https://bugs.documentfoundation.org/buglist.cgi?keywords=needsDevEval
+ New statistic tooling in place, runs every night
+ Found several assigned easyhack with status != ASSIGN, corrected
+ Started to review all easyhacks (slowly)
+ Interesting merge ratio gerrit(6112)-openhub(18239) so 2 of 3 are not in gerrit
+ good to track this ratio over time (Michael)
+ that's 1 year running - last 3 months should be better (Norbert)
+ would like to enforce the gerrit workflow for the release branch
+ do you mean require CI or sign-off ? (Miklos)
+ two ways to sign-off, cherry-pick locally, or push via gerrit.
+ concern wrt. the gerrit build latency (Michael)
+ just submit it two days earlier (Norbert)
+ current situation - release branches; (Stephan)
+ people push changes to gerrit, takes a while to sign-off
most of them jenkins already did a build.
+ what are we talking about ? (Bjoern)
+ little reason not to do it always.
+ if you push in your patch too late, if you don't
know when cloph tags it.
+ deep concern about process, over people (Michael)
+ have had long, deep queues in the past - 50 patches.
+ if this comes up - can we put it in
+ not discussing enforcing it in any way ? (Stephan)
+ two aspects (Norbert)
+ committers - not be able to push it bypassing gerrit
+ can still verified +1 and push it
+ want to see it built fine & overrule it (Norbert)
+ want the workflow to go via gerrit (Norbert)
+ and ideally let Jenkins verify it.
=> propose add an ACL rule for libreoffice-* - will have to push via gerrit (Norbert)
+ propose if bypass CI - you should be around during the build (Bjoern)
+ big batch of back-porting ? (Caolan)
+ no problem with the proposal.
+ can Jenkins prioritize release-builds vs. master ? (Michael S)
+ in theory - a plugin that does this (Norbert)
+ will look at it if there is a problem
+ Continued discussion on having a identifier for buildable master
+ an ongoing dev-list topic; need something ?
+ check the tinderbox (Cloph)
+ http://tinderbox.libreoffice.org/MASTER/status.html
+ all annotated with the gerrit hash
+ a tag or branch - kept updated ...
+ prefer to keep master buildable instead (Norbert)
+ new Jenkins hardware (Norbert)
+ expanding dbgutil and product builds as CI for master.
+ eventually build 6 - not 3x builds.
* Commit Access
* Developer Certification (Stephan/Bjoern/Kendy/Thorsten)
+ after GSOC setup
* Jenkins / CI update (Norbert)
ci at jenkins:~$ ./job_stats.pl 7
from:Thu Apr 14 14:08:09 2016
master linux rel jobs: 136 ok: 131 ko: 4 fail ratio: 2.94 % break: 4 broken duration: 0.75%
master linux dbg jobs: 118 ok: 113 ko: 5 fail ratio: 4.24 % break: 5 broken duration: 1.82%
master mac rel jobs: 130 ok: 129 ko: 0 fail ratio: 0.00 % break: 0 broken duration: 0.00%
master mac dbg jobs: 135 ok: 135 ko: 0 fail ratio: 0.00 % break: 0 broken duration: 0.00%
master win rel jobs: 114 ok: 111 ko: 3 fail ratio: 2.63 % break: 2 broken duration: 2.65%
master win dbg jobs: 108 ok: 104 ko: 4 fail ratio: 3.70 % break: 3 broken duration: 2.91%
master win64 dbg jobs: 113 ok: 105 ko: 7 fail ratio: 6.19 % break: 3 broken duration: 3.75%
lo-5.0 mac jobs: 5 ok: 5 ko: 0 fail ratio: 0.00 % break: 0 broken duration: 0.00%
lo-5.1 mac jobs: 19 ok: 15 ko: 4 fail ratio: 21.05 % break: 1 broken duration: 5.73%
+ should fail:
gerrit win rel jobs: 193 ok: 162 ko: 31 fail ratio:16.06%
gerrit lin rel jobs: 193 ok: 178 ko: 14 fail ratio: 7.25%
gerrit mac rel jobs: 193 ok: 181 ko: 12 fail ratio: 6.22%
gerrit all jobs: 193 ok: 152 ko: 40 fail ratio:20.73%
+ looking good, no breakage on Mac, some on Windows
+ one this morning gerrit didn't complain but tinderbox did.
+ 'always green' master-ness is in sight ...
* QA update (Bjoern)
+ UNCONFIRMED: ???
+ going up - help much appreciated.
https://bugs.documentfoundation.org/reports.cgi?product=LibreOffice&datasets=UNCONFIRMED
+ enhancements: ??
+ in ux-advise: ??
+ needAdvice at ??
+ regression analysis:
- We still carry a lot of regressions from the LibreOffice 4.2/4.3 releases (>50 each)
- Impress currently has a consistent low persistent regression rate (around 5 per release)
- Writer seems too have gotten a lot healthier recently (since 5.0)
- broadly we have 1/3 regressions in Writer, 1/3 in Calc and 1/3 is the rest
- roughly 20% of these regressions are marked as critical/major
+ punch-lines ?
+ if we keep new writer regressions low we'll do well (Bjoern)
+ we should write more regression / unit tests (Miklos)
+ Most Pressing Bugs:
http://tdf.io/mostressingbugs
+ tdf#89034 - OS X: Text not being justified in Persian and other RTL languages
+ Persists as Common issue -> lots of duplicates; many related layout bugs
+ Mail merge regressions (Bjoern)
+ Cor did some testing of the new code
+ http://tdf.io/mmregressions (11 open; 9 open last meeting)
+ several have commits on them - just need closing.
+ https://bugs.documentfoundation.org/show_bug.cgi?id=99258
+ cannot reproduce myself (cloph), but crash on startup would be bad...
+ non-release-config tinderbox is fine, release-style box shows the problem
+ April 22nd - 1st 5.2 bug-hunting session planned.
** This Friday **
* QA stats
+ https://bugs.documentfoundation.org/page.cgi?id=weekly-bug-summary.html
+149 -147 (+2 overall)
many thanks to the top bug squashers:
raal 17
Adolfo Jayme 12
Markus Mohrhard 12
mahfiaz 12
jan iversen 8
V Stuart Foote 7
Buovjaga 6
Michael Stahl 4
Jean-Baptiste Faure 4
Cor Nouws 4
Eike Rathke 3
steve -_- 3
Miklos Vajna 3
Ryan McCoskrie 2
Bryan Quigley 2
* Highest-Priority bugs (aka "MABs"):
5.2: 2/8 - 25% -1
5.1: 2/27 - 7% +0
5.0: 7/52 - 13% +0
4.4: 9/73 - 13% +0
4.3: 5/68 - 7% +0
4.2: 10/133 - 7% +0
4.1: 4/79 - 5% +0
4.0: 9/81 - 11% +0
old: 38/249 - 15% +0
* Bibisected bugs open: whiteboard 'bibisected'
+ 380/1334 383/1325 388/1313 391/1307 399/1305 404/1297 406/1292
+ http://tdf.io/bibibugs
* all bugs tagged with 'regression'
+ 748(-2) bugs open of 4848(+16) total 21(-2) high prio.
* ~Component count net * high severity regressions
Writer - 8 (+0)
Calc - 6 (-2)
LibreOffice - 3 (+0)
Base - 2 (+0)
Extensions - 1 (+0)
Impress - 1 (+0)
+ http://tdf.io/regressionscritical
* ~Component count net * all regressions
Writer - 234 (-4)
Calc - 147 (+1)
LibreOffice - 53 (+1)
Impress - 49 (+0)
Crashes - 44 (-2)
UI - 40 (+0)
graphics stack - 34 (+0)
Base - 34 (-1)
Borders - 31 (+0)
Draw - 26 (+0)
filters and storage - 22 (+0)
BASIC - 16 (-1)
Chart - 14 (+0)
Printing and PDF export- 10 (+0)
Formula Editor - 6 (+0)
Linguistic - 4 (+0)
Extensions - 3 (+0)
Installation - 2 (+0)
sdk - 1 (+0)
framework - 1 (+0)
+ http://tdf.io/allregressions
--
michael.meeks at collabora.com <><, Pseudo Engineer, itinerant idiot
More information about the LibreOffice
mailing list