minutes of ESC call ...

* Present:
    Hangout: Andras, Bjoern, Michael M, Thorsten, Tsahi,
              Miklos, Adam, Kohei
    Phone: Lionel, Eike, David, Norbert, Cloph, Petr,
            Pierre-Eric, Kendy, Tibbylickle

Trying etherpad for the 1st time: cf. unusual formatting.
    + http://pad.documentfoundation.org/

* Completed Action Items
    + poke at AFM startup / fontcache issue (Caolan)
    + invite pepp for next week (Michael)
    + working on Date API fix for rc3 (Eike/Michael S)
        [ made it into rc3 - many thanks to all concerned !
          Eike, Lionel, Michael S ]
    + sidebar option should default enabled it when enabled (Kendy)
    + sent vtune details to Pierre-Eric (Bjoern)
        [ pending response from Intel ]
    + talk with Robinson Ask => fdo integration / flow (Joel)
        [ doesn't particularly like moving all bugs -> ask
          may overload the admins & change our workflow,
          instead move ahead with feedback ]
    + get windows build-bot setup (Thorsten)
        [ one machine sent to Cloph, another doing
          it's first build, and nice howto from Kendy, which
          is here: https://wiki.documentfoundation.org/Development/Tinderbox/Setup ]

* Pending Action items:
    + get remaining GOSC guys introduced on the QA list (Cedric)
    + fdo#60858 needs input (Astron)
    + file a bug to split out the ruler-fix piece (Cedric)
    + file svg / help->about issue in master as 4.2 MAB (Astron)
    + improve configure to avoid versioning mistakes (Cloph)

* Release Engineering update (Cloph)
        + 3.6.7 rc2 update, prepared to be announced
    + 4.1.0 rc3 status ...
      + ready to be pushed to mirros after 3.6.7 has been done
        + pre-announce rc3 tomorrow
       + delays down to last minute iterator and indexing fix for Windows.
        + editing master-pages => release notes, have fix in master

* Should we kill TDF .deb builds ? (Bjoern)
  + pkgs are lame, neither Rene nor Bjoern working on that
  + direct people to Rene's builds / PPAs instead ?
  + long run: fix up-stream pkging to be more distro-like
  + significant differences Debian -> TDF (Lionel)
    + pwrt. internal libraries / embedded copies
    + nice to have a comparison to see where the bug is
    + TDF builds run anywhere; Debian/Ubuntu builds not as much
  + what is the real problem ? (Michael)
    + someone wants to install TDF packages, and uninstall them (Bjoern)
    + could we add a provide + a conflict (Michael)
    + people try to un-install the distro package - but remove
      the meta-package, de-install the meta-package
      (or forget pkgs like uno/ure which are not globbed by libreoffice*)
    + problems are inevitable for ISVs (Lionel)
  + just fixed 2x problems with Debian dependencies (Petr)
  + Stephan fixed one before vacation (Michael)
  + instead chance the pkging - to use a single name just
    libreoffice+ver everywhere (Petr)
      + get rid of ugly core-01 etc. pkgs - many of them with no ideas
        of what is inside them
AI: come up with proposal for what to do (Rene, Bjoern, Petr)

* QA update (Joel)
    + did the auto-feedback work for the next round of needinfo bugs
      + not closed yet, wait a month
    + counting almost done for the contest, so we should know winners soon.
    + feedback bits (thanks Joren for helping with this)
      + Robinson kindly committed to do the initial work
        + want it as simple as possible to start with, no twitter
        + happy + sad face etc. for sending general feedback.
        + a 'get involved' button.
    + button on the 'ask' site for 'report a bug'
      + admin button to move an ask question to FDO automatically.
      + ask admins are stretched already
      + let users pre-sort bug vs. ask etc.
    + Need_Advice bugs input appreciated (more of them now)
        + 10 bugs (+0) in that state (thanks)
        + http://tinyurl.com/ce6d8km
    + https://bugs.freedesktop.org/page.cgi?id=weekly-bug-summary.html
        +113 -95 (+18 overall)
        many thanks to the top five:
            ign_christian 13
            David Tardon 13
            James 6
            Caolán McNamara 5
            Michael Stahl 5

* round-trip interop. work (Tsahi/Adam/Miklos)
    + round-trip interop. of OOXML documents very important to us (Tsahi)
    + focusing on preserving data, to remove any fear of data loss
    + plan is to add missing data elements to the internal model
      + initially hidden, without rendering - just import/export
      + then incrementally add editing.
    + needed to add various MSO properties, or MCE / alternates
    + considering adding them to the data model.
    + currently several commits: some approved / some blocked
    + concern is around ODF; should we serialize some of this mess there ?
      + and pay a significant cost for doing that ?
    + some concerns (Miklos)
      + in general new elements in the data model / UNO API, we always
        updated the ODF filters too.
      + for DOCX filter we have to change the UNO API - because it is a pure
        UNO component.
      + one compromise might be do-able:
        add a special property that is a container for these; add a:
        sequence<string,Any> - and document in the IDL: "anything here will
        not be serialized to ODF"
      + perhaps we can live with that.
    + can we abuse the UNO tunnel for this (Eike)
      + sw already depends on writerfilter
    + need to discard the foreign properties on a object when it is mutated
      or work harder to determine what to do (Michael)
    + need to add a new property to each object ? (Adam)
      + yes, so - add a single new property 'InteropGrabBag' (Miklos)
      + with this container in the core, wouldn't have to touch the core much
        except to invalidate it
AI: + write a trivial example for paragraph properties (Miklos)
      + invalidate on copy/mutate - and load/save to OOXML
    + is the main motivation the smart-art, or other things ? (Kendy)
      + nope - it's everything (Michael)
      + also interested in correct rendering, but preserve first (Tsahi)
    + really sequence of sequences<string,any>

* Easy hacks (Bjoern)
    + fdo#62192
        + need alerts / annotations for attached patches ?
        + https://bugs.freedesktop.org/buglist.cgi?cmdtype=dorem&remaction=run&namedcmd=LibreOffice%20Bi-bisected&sharer_id=13921&list_id=323242
      + ask Cloph to provide some input on patches that are not responded to
      + have 43/339 Easy Hacks without action for > 1 year, should we review them ?
        + https://bugs.freedesktop.org/buglist.cgi?f1=days_elapsed&list_id=323253&o1=greaterthan&resolution=---&status_whiteboard_type=allwordssubstr&query_format=advanced&status_whiteboard=Easyhack&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&v1=365&product=LibreOffice
      + => discuss next week
      + what's the issue with old ones (Lionel)
        + concern around 'easy to fix' → but not fixed yet, users get upset (Joel)
        + not interested in upset users who complain about lack of trivial fixes (Michael)
      + concern wrt. lack of sexiness of easy hacks (Bjoern)
        + don't we just need to rotate the front-page of easy hacks ? (Michael)
AI: + rotate easy hacks on https://wiki.documentfoundation.org/Development/Easy_Hacks (Joel)

* Integrating Chrome's performance mgmt framework (Pierre-Eric)
  + have integrated it, quite simple, would like to merge it up-stream
  + http://www.altdevblogaday.com/2012/08/21/using-chrometracing-to-view-your-inline-profiling-data/
  + https://lh3.googleusercontent.com/mvBDUbV7r0zDfsFCMJ9m_C62ysvzgkD0BITuO1C94AvCn3WiBH_s-VcXggVAhLJXG3KnQ06z6gTztvKbzasFcXUgBBLO9BZZu_kWivjEjdrUiSm6lgQ
  + useful / can observe async events
    http://user.lanedo.com/~pepp/lo.json
    https://bugs.freedesktop.org/show_bug.cgi?id=39179
  + would like to ask the user to run the tool to get more detail on what's up
  + can use simple macros, re-use Chrome's ? with a small set of them, we can use them everywhere
  + a menu entry for start / stop the profile (?)
    + the JSON files have to be rendered by Chromium: chrome://tracing/
  + very small piece of C++ code
    + add trace-points around the code, and measure times between them.
    + begin/end event pairs around the place
    + async start + async end: start a timer, end a timer etc.
  + what is the run-time cost (Norbert)
    + check one boolean at the trace-points (Pierre-Eric)
    + and light footprint size-wize ? (Norbert)
      + two strings, category + name + thread-id + event type
      + can add JSON meta-data too if you want.
  + prefer to add it under debug build at first (Norbert)
    + if people like it, make it more widely available
    + timing in debug code is very different (Pierre-Eric)
AI: + polish & push to a feature branch (Pierre-Eric)
  + had this rtl_logfile in the past that did timings (Michael)
    + we're killing this (Bjoern, Norbert)

* GSOC update (Cedric)
    + should start merging student's changes into master around now
        + don't want a huge rush just before the mid-term
      + after the merge - continue to work on master ? (Norbert)
        + depends on the student & their preference (Kendy)
        + would like to encourage working on master (Norbert)
      + should encourage students to start working on master
    + mid-term is 29th July - until end of the week.
    + encourage students to blog etc.

* UX input (Astron)

* Certification Program (Stephan/Kendy/Bjoern)

* Coverity interview
  + in Markus' absence volunteer Norbert
    + AI (done) reply to coverity about accepting their request. (Norbert)

  + had a couple of requests for access, if not someone known:
    an active committer, denied access until more of a track record.
    + seems to make sense.

* IAccessible2 work
  + small & well contained: under 10k LOC
  + continued concern from around the place re: odd headers
  + should be compile-time conditional until that is solved
  + all code must have a clear LGPLv3 or ALv2 license & origin details in commit
  + should retain the old Java code until above mess is sorted out.
  + help merging much appreciated.

* Gerrit migration (Bjoern)
  + infra guys put gerrit on another machine
  + seems to be working, except that groups are not there
    + migrate now - no-one could push to master, except admins
    + need to fix this before migration.
  + other missing bits:
    + theming, gitweb can be missed for a few days, buildbot [!]
    + need to sort it out.
  + not involved in gerrit migration (Cloph)
  + suspect that data is stored with gerrit itself (Norbert)
AI: + will take a look at it (Norbert)

* Open 4.2 MAB
    + 1/1
    + https://bugs.freedesktop.org/show_bug.cgi?id=65675

* Open 4.1 MAB
        + 3 (of 55) older 7/52 7/48 2/42 4/39 5/37 7/32 3/26 3/23 4/17
             5% 14% 15%
        + https://bugs.freedesktop.org/showdependencytree.cgi?id=60270&hide_resolved=1

* Open 4.0 MAB
        + 12 (of 137) older 12/135 12/134 12/133 12/132 16/132 14/130 16/129
             9% 9% 9% 9% 9% 12% 11% 13%
        + https://bugs.freedesktop.org/showdependencytree.cgi?id=54157&hide_resolved=1

* 3.6 most annoying bugs ...
        + 60 (of 243) older 62/243 60/241 57/238 63/237 65/236 68/235 67/232
             25% 25% 25% 24% 27% 28% 29% 30%
        + https://bugs.freedesktop.org/showdependencytree.cgi?id=44446&hide_resolved=1

* Bibisected bugs open: whiteboard 'bibsected'
        + 34 (of 124) older 34/123 36/122 38/119 37/114 39/114 41/112 36/105 35/100
                http://bit.ly/VQfF3Q

* all bugs tagged with 'regression'
    + 287(-11) bugs open of 1740(+17) total

    * ~Component count net *
            Writer - 92 (+0)
       Libreoffice - 28 (+1)
      Presentation - 25 (-1)
          Database - 24 (-2)
       Spreadsheet - 24 (-3)
           Crashes - 21 (-1)
           Drawing - 16 (-2)
           Borders - 13 (-1)
             BASIC - 3 (+0)
         Migration - 3 (+0)

        + https://bugs.freedesktop.org/buglist.cgi?keywords=regression%2C%20&keywords_type=allwords&resolution=---&query_format=advanced&product=LibreOffice&list_id=36764
        + Migration: https://bugs.freedesktop.org/showdependencytree.cgi?id=43489&hide_resolved=1

Michael Meeks wrote (18-07-13 17:35)

* Should we kill TDF .deb builds ? (Bjoern)

For my understanding, that means that people that for reasons prefer to install without system integration and do testing and or work with their own deployment plan, ...
would not be able to download the versionXYZ_x86_deb.tag.gz any more from libreoffice.org/download ?
Handy alternatives ?

Thanks,

The proposal was to build straightforward tarballs instead of something
faking to be a debian package. That would still allow all of the above. Note
though that at least for Ubuntu the PPAs are really quick and allow you to keep
control of your deployment plan for major releases (e.g. the libreoffice-4-0
ppa will not update you to LibreOffice 4.1).

Best,

Bjoern

Hi Bjoern,

Bjoern Michaelsen wrote (24-07-13 12:07)

The proposal was to build straightforward tarballs instead of something
faking to be a debian package.

I should look up what a 'faked debian package' is :slight_smile:

That would still allow all of the above.

Using Alien I guess?
Note that in our very early days (hmm, you were not here at that time :wink: ) people were looking for 'debs' to download.
But if a workflow with alien is available by default, reasonable fast and good to explain... ok.

Note
though that at least for Ubuntu the PPAs are really quick and allow you to keep
control of your deployment plan for major releases (e.g. the libreoffice-4-0
ppa will not update you to LibreOffice 4.1).

Great of course, but that will only cover part of the needs & wishes I guess.

Cheers,
Cor