minutes of ESC call ...

* Present
    + David, Bjoern, Michael M, Andras, Norbert, Kendy, Markus,
      Miklos, Robinson, Stephan, Jan-Marek , Lionel, Cloph, Caolan,
      Jacobo, Eike, Michael S, Muthu
 
* Completed Action Items:
    + setup call to co-ordinate new load/save/validate hardware (Markus)
    + stick with the GetSolarVersion plan & put it into the release notes (Marcos)
    + revert boost::rational changes for now (Kendy)
    + poke Taras wrt. UI / telemetry code re-use here (Michael)
 
* Pending Action Items:
    + get a clear use-case & description of what doesn't work vs. master wrt. Calc sorting (JBF)
       [ Work in progress. At the moment I have at least 3 different use-cases:
          # bank account: the sorted range contains a column where each cell contains the same 
formula (fdo#81633 and fdo#85215)
          # sorting students by their scores: the cells in the sorted column contains formula with 
functions like AVERAGE() or SUM() (fdo#79441)
          # computing an indicator from data: the cell containing the formula with references is 
outside the range of cells that is sorted (fdo#45146)
          # I am pretty sure I will find other use-cases ]
AI:    [ poke JBF - what is the status here ? and what to do next ? ]
    + look at regressions associated with a SUSE bugzilla bugfix (Timar)
 
* Release Engineering update (Cloph)
    + 4.2.7 & 4.3.3 released as final today
AI:     + mention calc sorting issue in the 4.2.7 release notes.
        + unclear what went wrong wrt. bug reporting / fixing / on
          that issue and/or how to fix it.
          + Questionable new feature breaks existing workflow / new behavior set as default / new 
feature backported to stable 
    + 4.4 Alpha 1 status
        + horrible problem with calc - not refreshing the background
            + zoom issues etc. a display problem.
            + fixed via a revert.
        + problem building all languages on Windows
            + prolly the PATH limit too: for now building in c:\lo\
            + working on a fix to make it less reliant on v. short paths
        + how did the VS 2013 build go
            + no problems related to this identified yet;
            + do we get issues with earlier CPUs with no SSE2 ?
        + good that we discovered these breakages early (Bjoern)
AI:     + poke at fdo#84854 (Michael)
            + no harm in reverting for now, can always replace.
    + 4.4 Alpha 2
         + planning to do a new tag later this week for testing.
    + 4.4.0 Beta 1 / Feature Freeze
         + due November 17th.
    + 4.3.4 RC1 - Nov 24th
    + Android Remote
 
* Crashtest futures / automated test scripts (Markus)
    + call on Tuesday; new testing hardware.
    + result - get a Manitu server & leave room in the budget for
      ondemand Amazon instances (with spot pricing) if there is
      special need at some point.
    + will migrate when we have the perf. crash. testing & fuzzing,
      and should bring down all the times.
         + perf testing: in 2-3 hours (same day)
         + crash testing: to ~1 day.
    + talked with Matus: a way to notify developers similar to tinderbox.
    + could have bibsect repositories there too (Norbert)
    + and at least 1 VM to try crazy new testing ideas.
    + budget specifically for load/save/validate testing (Michael)
        + any concerns / Bjoern ?
        + lets not let that creep in wildly; test new things,
          but not burn lots of CPU.
    + what was missing - a real comparison of prices ? (Bjoern)
        + comparison happened on the mailing list by now.
    + decision to buy the base-load and dynamically rent the discretionary work.
    + when does it arrive ? (Michael)
        + Florian pinging Manitu wrt. conditions etc.
        + will take 2-4 weeks to get it setup.
    + right after Beta 1-2 will that be a good time ? (Bjoern)
        + mostly just migrating the VM(s) across (Cloph)
        + ~trivial to migrate the workload (Markus)
 
* Crashtest update (Markus)
    + http://dev-builds.libreoffice.org/crashtest/013ec10d4e89e86cbd53a88058defb7e97ebaa49/ 
(25-Oct-2014)
    + still a big spike in DOCX export crashers
        + if no-one beats him: will look into it next week (Miklos)
    + otherwise similar to last week
        + most of the PDF import crashers fixed since last week
    + RTF import & export are now set to zero (Miklos)
 
* Seattle Hackfest (Robinson/Norbert)
    + around 32 people showed up
        + lots of Linux & Mac users showed up, few Windows users
        + lots of interest in getting involved.
        + several people eager to come but couldn't make it - eager to come next time.
    + lots of build work (Norbert)
        + lots of setup / long & heavy etc.
        + got 5-6 new people with builds
    + how did pre-built VM's fare (Michael)
        + a couple of guys used that (Norbert)
        + 1x person on IRC, asked for help with x2go connection (Cloph)
        + saw 1x building & 1x using only.
        + better documentation - good for next time ? (Robinson)
    + VS 2013 express user (Norbert)
        + ran into a problem wrt. wrong SDK; apparently a
          blank Windows 8.1 with VS.Express is not good.
    + badly need ways to stop 'building' consuming the hack-fest (Michael)
        + or a way to keep hand-holding after the hack-fest (Bjoern)
    + lots of people expected to use a VM for build (Kohei)
        + lots of questions, but couldn't help them.
        + many ended up using their machines to build,
          because remote / Amazon VM wasn't setup.
        + students machines, not powerful to build in linear time.
        + getting details of VM's in advance & getting trained
          on these is critical.
        + vast majority of people wanted access to VM
        + monster communication problem with cloph ...
            + Robinson was briefed on how to do that (cloph)
                + Info emailed to hackfest devs (Robinson)
            -> communication droppeded <-
AI:     + work out how to use VM's (ssh bob@1.2.3.4) (all hack-fest mentors)
            + assume next time that all participants need Amazon access (Kohei)
                + students are unfamiliar with the time to compile etc.
    + time to clone the repo - took significant time (Norbert)
        + great / luxury problem to have - lots of attendees (Bjoern)
      https://wiki.documentfoundation.org/Events/2014/Seattle_LibreFest
    + would be nice to have some blog-posts / picture posts (Bjoern)
    + would it be better to have a chroot pre-populated ? (jmux)
        + problem not setting up the env. - but building (Bjoern)
        + containing the whole tree / ccache etc. (jmux)
             + a setup change & try things out
        + do need to get them working locally anyway (Michael)
             + can't afford the Amazon VMs indefinitely.
        + could create chroots of builds as we release (Bjoern)
        + reasons built in VirtualBox (Muthu)
             + copy that across & concentrate on hacking instead.
             + VirtualBox runs on Win / Mac too (Michael)
    + Cloph has a kick-start file for the full build env.
        + is it possible to ask releng to build those ? (Michael)
        + icecream can work in the cloud ? (Norbert)
             + not ideal in a hackfest - transferring Gb's (Moggi)
 
* Hackfests
    + Toulouse Hackfest (Arnaud)
        + Confirmed for Nov. 15-16th
        + more details: https://wiki.documentfoundation.org/Hackfest/Toulouse2014
        + budget reserved (Eur 2k) - poke Bjoern.
    + Munich hack-fest (Jan-Marek):
        http://www.it-muenchen-blog.de/2014/07/bug-squashing-party-2014/
        + jmux (IRC) can add you if you don't want to create an account
        + will have a BSP come-together; if someone says they're coming
          just show up etc.
           + November 21st-23rd, 2014
             https://wiki.debian.org/BSP/2014/11/de/Munich
           + we should drop by with 2 or 3 developers, if possible
    + Need further offers [!] ...
        + propose FOSDEM (Michael/Moggi)
            + and try to get everybody there; Kohei etc.
            + add days after it (to avoid slide overrun)
            + also the month before the release (Bjoern)
            + 31 January & 1 February 2015
        + propose Collabora / Cambridge (Michael)
        + should consider Las Palmas de Gran Canaria/University again (Bjoern)
        + Freiburg - also an option (prolly better in summer anyway (Bjoern)
 
* UX Update (Kendy)
    + wasn't there - but the call took place
    + also paragraph background; voted against removing b/g colour
    + general results from copy/cut/paste survey: not clear yet.
       + discuss at the next call.
    + default option of sorting / reference update (Kohei)
       + query UX there
       + can we get input there.
       + so in master / 4.4 it's just a officecfg setting (Michael)
           + same in 4.3 - 4.3 default is old-style.
       + discuss it in the next UX call (Kendy)
AI: + push the new default to libreoffice-4-2 branch ? (Andras)
       + despite the lack of a release vehicle
           + ack (Eike)
 
* Certification (Stephan/Bjoern/Kendy)
    + consensus approval of new certification candidates
       + happy for them all to have commit access
    + various others in the pipeline ...
       + if you want to get certified - ping Kendy, Stephan, Bjoern
    + are we encouraging people to apply for membership ? (Michael)
       + yes, routinely when we invite for certification and they're not yet members
    + reviewed all expired certifications; will renew all.
  
 * Google Code-In  (Robinson)
    + https://www.google-melange.com/gci/homepage/google/gci2014
       + anyone want to submit an application by November 10th ?
       + sadly deadline by FF & schedule through bug-fixing (Michael)
 
* Bugzilla migration - discuss next time (Bjoern)
 
* QA (Robinson)
    + UNCONFIRMED bug count down to 645
        + big thanks to all our triagers !
        + couple of new people actively helping
        + some concerns wrt. regressions creeping up.
            + this tends to happen when we shrink unconfirmed a lot (Michael)
            + when a bug becomes verified - some will be regression (Norbert)
            + high currently, FF in 3 weeks (Michael)
 
* Interesting QA stats as ratios (Bjoern)
    + bugging him that Writer was leading; is it truly more regression prone ?
        + looked at stats instead as proportion of total bugs filed:
        + more popular components have more regressions
        + all ratios vs. count of confirmed bugs that are open.
 
    These are the components with the highest relative regression ratio:
 
        graphics stack    39%
        extensions        14%
        files and storage 11%
        database          10%
        printing and PDF  10%
        spreadsheet       6%
        draw              6%
        l10n              6%
        chart             5%
        writer            5%
        ...
 
     So writer is not having more regressions than the rest, but likely
     just a bigger userbase (or contain more functionality).
 
     also looked at which components have the most enhancement requests:
 
       impress remote  79%
       ux-advise       72%
       l10n            56%
       UI              42%
       linguistic      34%
       WWW             34%
       formuala editor 33%
       chart           32%
       database        32%
       LibreOffice     32%
       ...
 
     Also interesting are the highest unconfirmed ratios, showing 
     either more users than QA or a high presence of heisenbugs:
 
      sdk              86%
      impress remote   43%
      installation     34%
      ux-advise        31%
      BASIC            26%
      extensions       25%
      framework        25%
      printing and PDF 23%
      LibreOffice      18%
      Linguistic       16%
      ...
 
* QA stats:
  + https://bugs.freedesktop.org/page.cgi?id=weekly-bug-summary.html
    +143    -119        (+24 overall)
    many thanks to the top bug squashers:
        tommy27             16
        Beluga              10
        Maxim Monastirsky    9
        Adolfo Jayme         8
        Jay Philips          6
        Miklos Vajna         6
        Jean-Baptiste Faure  4
 
* Open 4.4 MAB
  + 3/16 4/16 5/15 4/14 3/12 3/12 3/12 7/12 3/8 3/8 4/8 4/8 3/7 3/7 2/5
  + https://bugs.freedesktop.org/showdependencytree.cgi?id=79641&hide_resolved=1
 
* Open 4.3 MAB
  + 16/64 15/63 16/63 16/60 17/61 16/60 14/59 15/58 12/54 13/53 14/52 14/50
     18%   23%   25%   26%   27%   26%   23%   25%   22%   24%   26%   28%
  + https://bugs.freedesktop.org/showdependencytree.cgi?id=75025&hide_resolved=1
 
* Open 4.2 MAB
  + 84/262 82/260 80/260 82/260 86/257 88/257 89/257 91/257 86/252 82/245
     32%    31%    30%    31%    33%    35%    34%    35%    34%    33%
  + https://bugs.freedesktop.org/showdependencytree.cgi?id=65675&hide_resolved=1
 
* Bibisected bugs open: whiteboard 'bibsected'
  + 170/441 161/427 153/413 154/410 147/401 143/392 143/389 146/385
    + http://bit.ly/VQfF3Q
 
* all bugs tagged with 'regression'
    + 496(+8) bugs open of 3165(+36) total
        * ~Component   count net *
                Writer - 147 (+1)
           Spreadsheet - 101 (+7)
          Presentation -  34 (+4)
               Crashes -  33 (+3)
              Database -  28 (+0)
                    UI -  25 (+2)
           Libreoffice -  24 (+0)
               Drawing -  23 (-1)
              Graphics -  22 (+0)
               Borders -  19 (-2)
               Filters -  18 (+0)
           Print / PDF -  14 (+0)
                 Chart -  12 (+0)
          Installation -   5 (+1)
            Extensions -   5 (+0)
            Linguistic -   4 (+0)
                 BASIC -   4 (+0)
               Formula -   3 (-1)
             framework -   1 (+0)
                   sdk -   0 (+0)
        Impress Remote -   0 (+0)
  + http://bit.ly/15mM2Yn - for devs ( no NEEDINFO / UNCONFIRMED )

-- 
 michael.meeks@collabora.com  <><, Pseudo Engineer, itinerant idiot