minutes of ESC call ...
michael.meeks at collabora.com
Thu Mar 27 08:51:53 PDT 2014
Bjoern, Eike, Norbert, Christian, Adam, Matus, Kohei, Kendy,
Cedric, Miklos, Andras, Michael M, Stephan B, David, Robinson,
* Completed Action Items
+ unwind non-closed 4.0 MABs (Michael)
+ send bluez 5.0 bug # to Caolan (Michael)
[ ahunt looking into this ]
+ update certification website (Kendy)
+ create a meta-bug for large l10n relevant 4.3 issues (Andras)
* Pending Action Items:
+ crashtest box - more research (Moggi/Cloph)
[ still pending discovering which document is the one ...
box is doing something now (Cloph) ... ]
+ will try filing a lot of small string changes as easy hacks (Astron)
* Release Engineering update (Christian)
+ iOS remote approved with fix for iOS5
+ will pull the Android remote translations, and will publish
that as a public release
+ 4.2.3 rc2 status
+ pushed to early testing & QA already, synching to the mirrors
+ will be released today / tomorrow morning.
+ on-line update is triggered by micro version being larger
+ so far only 4.2 getting updates to 4.2.2
+ users on 4.1.5 etc. not getting updates to that.
+ 4.1.next - April 7th
* UX update (Astron)
+ not much to report
* Daylight saving reminder:
+ same European local-time next week
* Performance regression testing (Matus)
+ raw data from Matus' callgrind / tinderbox profiling script is:
+ recently expanded to use around 10x test files
+ we load them, and save them and log the pseudo-cycles
+ the hope is that this will run regularly (taking ~10 hours)
and produce a long time series to spot performance problems.
+ concerned by the big 7x XLSX test jump (Norbert)
+ in beginning, only loading, but now exporting too, and
the export takes a lot longer.
+ the big drop is 20x days of data we lost
+ performance seems stable, if not very slightly better over time
+ what do the numbers mean ? (Kohei)
+ they are callgrind pseudo-cycles - using cache simulation (Matus)
+ can we publish the latest callgrind output files ? (Michael)
+ sounds reasonable
AI: + look into automatic up-load of data & last callgrind traces (Matus/Cloph)
+ where is the script ? (Michael)
+ in buildbot.git/loperf/*
+ test files in testfiles.git/loperf
+ why are numbers in history.fods different ? (Kendy)
+ divided by a million to make it easier to read (Matus)
+ also measuring the first-ever-start and 1st start (Matus)
* Security-related bugs in release notes (Michael Stahl)
+ users should not be adding those, eg:
+ not really a security bug & => should be removed.
+ shouldn't list security bugs in the release notes (Norbert)
+ many bugs are security bugs
+ concerned wrt. the process too (Bjoern)
+ if we do fix security bugs we have the page here:
+ mozilla have a single CVE for each release (Michael)
+ that includes all potential security related issues
+ concern with other projects using the code
+ concern wrt. total Mitre non-responsiveness
+ leave as-is.
AI: + remove recently added, misleading security bug section from release notes (Michael S)
* GSoC update (Cedric)
+ have around a dozen good looking students
+ final slot request due April 7th
+ requested 9 min, 12 max.
+ continue rating the students in the meantime at:
* New committers (welcome to the team) (Norbert)
+ Chris Sherlock - great work in VCL
+ Maxim Monastirsky - impressive calc work (Kohei)
* Crashtest update (Markus)
+ pending the fix.
* Gran Canaria hack-fest (Bjoern)
+ April 24-25, 2014 (Thu-Fri)
+ book flights etc.
+ hotel is booked by TDF from Wednesday to Saturday see
https://wiki.documentfoundation.org/Hackfest/GranCanaria2014#Accomodation for details
+ venue is online too: https://wiki.documentfoundation.org/Hackfest/GranCanaria2014#Venue
+ bus from the hotel -> university planned.
* Certification Committee (Stephan/Bjoern/Kendy)
AI: + inform marketing guys wrt. certification (Kendy)
* UX test at LibrePlanet / SpinachCon (Robinson)
+ several bugs / rendering issues to enter
+ events are important, get good feedback etc.
* QA (Robinson)
+ doing a bit better with bug triage this week
+ stable at 930 or so unconfirmed
+ almost no 4.2 writer regressions
+ down to about three
+ asked QA veterans to priotize old bugs and
identify which are real regressions.
+ buzilla migration - not happening
+ crawling fdo bug XML with
+ hope to have whole state of fdo bugs locally by then.
+ possibly other targets for that ...
+ need to merge bibisect repos, update bibisect docs (Bjoern taking over)
+ likely needs more space on dev-downloads, but infra is on that AFAIK
* QA stats:
+153 -144 (+9 overall)
many thanks to the top bug squashers:
Björn Michaelsen 9
Andras Timar 7
Adolfo Jayme 7
Caolán McNamara 5
Julien Nabet 4
Kohei Yoshida 3
Stephan Bergmann 3
Markus Mohrhard 3
* Open 4.3 MAB
+ 0/1 1/1 1/1 1/1
* Open 4.2 MAB
+ 20/112 19/111 20/111 18/104 17/102 17/99 18/97 19/95 18/85 17/70
17% 17% 18% 17% 16% 17% 18% 20% 21% 24%
* Open 4.1 MAB
+ 79/206 84/206 85/206 86/204 81/198 80/197 37/149 37/149 34/144
38% 40% 41% 42% 40% 40% 24% 24% 23%
* Open 4.0 MAB
+ all migrated to 4.1
* Bibisected bugs open: whiteboard 'bibsected'
+ 38/185 40/185 39/182 39/181 40/180 40/180 42/180 42/179 43/178
* all bugs tagged with 'regression'
+ 333(-6) bugs open of 2427(+13) total
* ~Component count net *
Writer - 87 (-3)
Spreadsheet - 57 (+3)
Presentation - 30 (-2)
Libreoffice - 26 (-2)
Database - 25 (+1)
Crashes - 21 (-1)
Drawing - 13 (+0)
Borders - 13 (+0)
Migration - 3 (+0)
BASIC - 3 (-2)
+ http://bit.ly/15mM2Yn - for devs ( no NEEDINFO / UNCONFIRMED )
michael.meeks at collabora.com <><, Pseudo Engineer, itinerant idiot
More information about the LibreOffice