janneke.blog

Welcome to my blog. Having managed to avoid the [b]logosphere for quite some time now...now let us find out where this goes.
You can also burn useful hacking time by having a look at the logs from my GNU/Novell/... friends

• Michael Meeks
• Federico Mena-Quintero
• Han-Wen Nienhuys
• Tiggr
• Connie de Vos
• [Telsa Gwynn]

2008-04-26

• Sleeping on yesterday's questions about the slow pace of maturing of our community prompted the feeling that it's time for me to re-read Kevin Kelly's Out of Control: The New Biology of Machines, Social Systems, & the Economic World. How many successful examples of transformation into an open source community do we have? It took Netscape quite a while to become Firefox and that may not be the path that Sun is seeking to follow at this time.

2008-04-25

• Most people interested in OOo or Sun's OpenSource struggle will have found Ted Ts'o sun-solaris blog entry on /. That's how I found yesterday's entry on organic-vs-non-organic where he writes

  Solaris, of course, is doing something else quite different, though. They are trying to make the transition from a proprietary customer/supplier relationship to trying to develop an Open Source community (..)

  We don’t have a lot of precedent for projects who try to go in this direction, but I suspect they are skipping a step when they try to go to the end step without bothering to try to make themselves open to outside developers. And by continuing to act like a corporation, they end up shooting themselves in the foot.

  A nice wording of the sentiment in the FLOSS community. I have been wondering for quite some time now how it is possible that even small projects like LilyPond and GRUB 2 seem to have a much more lively user and developer community than OpenOffice.org has. Is office software so much more unsexy than music notation or bootloading? Is OpenOffice.org so much more bug-free that users/developers don't need to get involved? Are we as developers not taking enough pride in our work? Is OpenOffice still too young? Is there not enough competition, too little eyeballs? Are we hurt by artificial (corporate) agenda's or a cultures that prevent writing of beautiful code?
  So how is this transition from being a corporate project which happens to publish most source code to a true organic open source project supposed to happen? How can we speed this process up, or even prevent it from being slowed down?
  Should we seek to divide and make work of our Go-OO.org community: we fix getting your patches in and deliver a buildable tree? It is doubtfult if that is what would serve OpenOffice best. But what can we do? What are people at Sun thinking, are they seeing what we are seeing? Is there any guidance in this or is there too much guidance, are there expectations or dreams, is there a plan?
• Cleanups for building and running sortdlg.
• Convert Sort dialog's Options page code to Layout, implement missing stuff. TODO: AddTabPage(..)

2008-04-24

• Use Window as base for plugin instead of Control.
• Prepare up to date patches for ooo-build.
• Look at using Layout for just one tab of an old .src dialog: get Calc's sort dialog to build and eventually run in the standalone test app. Thanks Kohei!

2008-04-23

• Happy to find two more zypper bugs I filed are fixed.

  #348669 - zypper should print progress to stderr  
  #348676 - zypper should list package descriptions/summaries

• Re-implemented plugin patch using VCLXContainer and HBox as base classes. This leads to the same detached result and moreover it confuses me: into what container add what incarnation of the Plugin?
• Poke at both implementations for a few hours and finally find that the [re]draw and detachment problems arise from the silly fact that at time of creation the DialControl Plugin's parent is 0. Apparently this does unfixable damage, at least not fixable by reparenting later. Anyway, the dialog now looks and moves fine.

2008-04-22

• Add plugin patches to ooo-build. Remove buildfix-xmloff-exportxforms patch from dev300.
• Fresh ooo-build with Factory.
• Look at calc shortcuts.
• Talk with Michael about layout plugins.
• Team meeting.

2008-04-21

• Visit Ei/PSI grand opening. Inspiring and sobering talks by Whitfield Diffie, Andrew Odlyzko, Ian Brown and Bruce Schneier. It is amazing to watch the vastness of the domain that people like Andrew and especially Bruce can choose to talk about, what stretches they make and how they just get away with it. No, that's not it. Great speakers understand the art of how talking about big and profoundly simple things, that could others easily cost their credibility and respect, makes them bigger.

2008-04-20

• Set out to make fresh pasta today. A couple of years ago I tried it for the first time, inspired by a lovely meal that Han-Wen prepared for us. It was a mixed succes then, some bits were really too thick. This time it went very well though. I learned again that big eggs are much bigger here than Marcella Harzan buys them in Italy. Although I tried to use the original method of stretching and flattening the pasta for a bit, in the end I used a common Dutch rolling pin and got a nice evenly thin sheet of pasta.

2008-04-18

• Poke some more at Plugin; prepare two nice sets of patches.
• Update to Factory. zypper dist-upgrade worked flawlessly, after manually upgrading rpm first.

2008-04-17

• Get plugged-in svx:DialControl to hello world through UnoTunnel. When the dialog pops up, the size is already right! The contents of my editor window show up as background, but no complaints for a first result.
  And it gets even better, clicking on the expected location shows a functioning DialControl. Woo!
  It doesn't get much better, though. Altough the DialControl is reparented using the ModalDialog from the UnoTunnel, they somehow act as separate windows. Spend rest of day poking these separate/unattached window, background and redraw issues.

• With my fresh zypper, try a dist-upgrade to factory.

2008-04-16

• Look at Florian's Windows keybindings doc.
• Look into custom widget plugin & unotunnel.
• Try to rebuild guile from CVS, pleased to find they moved to GIT. Guile now depends on gnulib, but that seems not to be included in openSUSE 10.3? Package gnulib, build guile. I'm wondering how Debian does it, to always just have packages for everything. Are all upstream developers using Debian and adding all their dependencies? Am I looking in all the wrong places, should I be adding some repositories? Or is this a culture thing, are SUSE hackers Gentoo/FreeBSD-users in disguise who just compile and install everything from source? Maybe I'm just unlucky today.
• Trying to install gcc-src I find that my zypper trunk dumps core. Rebuild zypper (libzypp, sat-solver) trunk. Install gcc-4.2.1 src (zypper source-install gcc42) but fail to find any references to visibility in the changelog. Hmm.

2008-04-15

• Find overnight build failed on win32-agfa patch. Fix it, svn up, conflict with Kohei's fix :-). Start build.
• Look into recover dialog integration.
• Support Eric Bachard (whom I also had the pleasure to meet in Prague) on layoutdialogs CWS build on Mac, get a build fix :-)
• Team meeting.
• What a nice coincidence, Corrie ten Boom's The Hiding Place on national tv.

2008-04-14

• Catch up email.
• Sort out hackings from laptop.
• Fix some bugs in layoutdialogs.
• Start build for exportForms / visibility thing.

2008-04-11

• Today is the first public/community day of GoOOCon. Improvise short talk and demos on layout dialogs status with Ricardo.
• Some great talks, esp. Michael's about old and new cruft in OOo and Julian's on Valgrind's (valgr[i^]nd) new memory and thread tools.
• Leave early. Some real bad luck/planning with the train connections, should have had 1h30 leeway but still managed to miss the last bus. Taxi home. Collapse.

2008-04-10

• Third day of GoOOCon.
• Brainstorm session on go-oo.org, CMT 1-1 with JP.
• Talk with Kohei and Michael about layout plugins, Michael as always sees no problem whatsoever and makes his thoughts clear in one line of code. That helps! Michael wonders if I have a working layout dialogs demo setup. I don't, both the upstream layoutdialogs CWS and ooo-build dev300 builds I prepared have problems. The ooo-build dev300 build has a new weird install structure openoffice.org/basic3.0 that our standalone test program barfs on, the upstream build suffers from a

  "missing vcl resource"

  and images are broken. Rush to checkout and download ooh-m12 before dinner, that should work.
• Another nice dinner, real big group this time, talking with Florian and Shaun. Build plain ooo-build trunk ooh-m12 to prepare for talk and finish layout dialogs status-writeup until late. Good move to put everything also in stable, it just works there :-) Just when it all works I get a heartbreakingly friendly SMS from Michael.

2008-04-09

• Second day of GoOOCon.
• Again a warm lunch today. Question kendy about this, he manages to make me believe the Chech always have a warm lunch, just like the Italians. Either way, it works for me this week.
• The lightning talks give a great view of what everyone's been doing and who everyone is. Noel used to live in Cork, Kohei and John live fairly close by but never meet, etc.
• Discuss Celss dialog with Kohei. Agree to hack-up a concrete rather than abstract implementation of a DialControl in the zoom dialog to increase understanding (hours, days weeks of work?) We get it running in an hour, so estimate ~1 day coding.

2008-04-08

• First day of GoOOCon. At breakfast, meet Fridrich. He's a real funny guy, even dryer in person than on IRC. Try to get back at him with a TrainedMonkey joke, but it falls flat in the telling.
• Team talks and breakout sessions and lots of time to talk with Kohei, Florian, Thorsten, Tor.
• For dinner, a great stone-oven pizza. I had worse in Italy.

2008-04-07

• Travel to GoOOCon. by plane takes me eight hours. Google says: 7:43 by car, but then you'd probably have to traverse Prague, so add an hour. International trains remain broken in Europe, they take 14:31 up till 18:43 hours and 7 transfers, reservation is mandatory.
• Using Michael's blazingly fast replied SMS directions find almost all the guys in a pub nearby. Just great to meet everyone, to have a face to go with the name or nick!

2008-04-02

• Extract latest layoutdialogs to ooo-build dev300 and test.
• Prepare laptop for building full OOo.

2008-04-01

• Work on layoutdialogs CWS with Christian. Lots of discussions today.
• Look into java properties. This would at least mean moving translation from compile time to runtime. Otoh, it would make the translation process more uniform with the rest of OOo. Now the question is, would that be all good or all bad?
• Team meeting.

2008-03-31

• Find example use of ZIP files in scp2 process. Play with it a bit and install xml file tree using ZIP archive. Thanks, Christian!
• Makefile updates, factor out a nice layout.mk for dialogs.

2008-03-28

• Add error box with printf if we cannot find any matching .xml file.
• Move installation and compilation of xml files to sw and svx modules.
• Christian hints about using ZIP archives to do the scp2 installation step. That would be nice, it would fix the current silly problem of mentioning every single translated xml file in the d.lst files and the scp2 install snippet.

2008-03-27

• Finish New Hire course.
• Work on layoutdialogs CWS with Christian. Everything seems to compile now on his side, but it does not run yet. Talk about the translation process and how tralay works. It's just grand having Christian work on this with me. It is so much easier, quicker and more pleasant than working through bug reports or EIS comments. Moreover, Christian knows a lot about OOo development, which really helps. I do wonder though what it would do for our developer community if we were to discuss all technical issues and especially have all patches pass through a public list.

2008-03-26

• Work with Christian on his/our Windows/Solaris/(and Linux!?) build problems.
• Do remaining New Hire Training, take CNS exam.

2008-03-25

• Update ooo-build.
• Catch up with New Hire Training.
• Team meeting.

2008-03-22

• Made time to read Corrie ten Boom's The Hiding Place that Michael gave me to read when I was their guest in Newmarket. A most inspiring story of amazing spiritual beauty. It even comes with three practical exercises (relating to the book) for the reader on how to use and grow spiritual strength. Corrie's exercises are interesting as they have elements of the Walk for Atonement, the Compassion exercise, Label V and perhaps a wise word-lesson from someone who has had a Source List experience: [try] to be here right now. Unless you are very creative maybe, Corrie's exercises do require a strong belief in God for them to work.

2008-03-21

• Look into recover dialog.
• Oops, add missing files to CWS and mail with Christian.
• Find I'm out of space, play with lvextend and resize2fs to make full use of my big disk.
• Find my layoutdialogs-cws build was skewed, do a full rebuild.

2008-03-20

• Implement MoreButton by simple reuse of AdvancedButton.
• Open new CWS on DEV300_m3: layoutdialogs and move all GIT patches and development there.
• Update Carsten Driesner and with Christian Lippka about layout development, invite them to send patches and build-fixes :-)

2008-03-19

• Continue m3-build.
• Fix recover dialog issues.
• After finding MoreButton in Michael's pivotfilter, simplify recover dialog to use new `AdvancedButton'-like concept.
• Extract `AdvancedButton'-like concept into new widget, use it in recover. Add empty MoreButton skeleton alongside it.
• Make new screenshots and add them to bug for review. And, oops, set bug to ACCEPTED.

2008-03-18

• Deliberately slow day.
• Finish CMT, take a nap.
• Chat with kendy about estimates.
• Talk with JP.
• Team meeting.
• Join EMEA quarterly meeting.
• Move layout diffs from m2 to m3 in preparation of making a CWS, start build. Hopefully m3 has less non-layout related build problems. (notice uselessness of CVS diff without commit-access wrt adding files :-)

2008-03-17

• Unproductive day. In the morning, worked on estimations and CMT.
• After lunch, took a nap and woke up at 17:00 with a 39C fever.

2008-03-14

• Make VBox, HBox vcl wrappers usable. Test with recover.cxx, but revert to using individual visibility instead because of spacing/sizing problem. Running the recovery dialog from DEV300-m2 has a pleasant surprise: the progress bar has been updated to use a more modern GNOME-like one.
• Find that FixedLine and zoom.cxx have been extended upstream. Implement FixedLine.IsEnabled () and add new View Layout functionality to zoom.xml; merge and create minimal diff for zoom.hxx, zoom.cxx.
  After running the upstream dialog, I realise that Zoom factor and View layout parts are in columns, alongside each other.

• After adding an extra vbox and hbox this is what the new layout-based zoom dialog looks like. Now I need to redance the whole translation hoopla for the new strings.

• Interesting call with jonp discussing (or should I say agreeing over ;-) Key Binding Compatibility estimates.

2008-03-13

• Get ProgressBar to display in recovery dialog (thanks Michael!).
• Implement FixedImage.
• Add Simple/Advanced mode to recovery dialog. I loosely based it on the design made by Sigi in bug #194434. Only loosely, because I'm not sure how to do the floating box with documents. But also because I am no particular fan of the hidden right-click functionality anyway, which then has to be documented and is also bothering the casual user. Do we also need to display the recovery status with each document?
• Got DEV300_m2 to build. Get some cxx files here and there from HEAD that have an obvious mismatch with their hxx's.

2008-03-12

• Various DEV300_m2 compile fixes. Upstream still does not compile with gcc's visibility stuff.
• Work on ProgressBar.
• Update Carsten Driesner about layout.

2008-03-11

• On with build fixes for DEV300_m2. Some libraries have not been renamed without $(VERSION) or $(UPD).
• Team meeting.
• Phone with Michael, decide against HBox/VBox split. This would add some clarity but lead to code duplication. While the code is a bit cryptic to read, it is really tight and well written by Ricardo.
• H. phones to tell me, quite excited and proud, that she amazed and inspired her occupational health coach to go on and take the Avatar Master course. So they'll be going to Egmond together in April! She also amazed her boss, he will be paying for her Masters too.

2008-03-10

• Back to another passion of mine: free software in the form of OpenOffice.org. Catch up email and update to DEV300_m2.
• Look some more into vbox/hbox thing: refactor and simplify HBox's calculateSize/getOptimalSize.

2008-03-09

• On with the rundowns. W. has another breakthrough with the persistent mass rundown. After lunch we get to hear the recording of Harry's happy tilapi talk. My favourite quote: You can never get enough of the things you don't need to be happy.
• All of our students finish the course. Everyone has done very well. Thanks to all for the work done in consciousness. Thanks for all the fun. Thanks to the aligned team! I'm grateful for having been part of this. This international course has been the biggest one we ever held in Europe.
• Back home at 23:00, T. stayed up for me. I enjoy watching A. and B. so sweet asleep.

2008-03-08

• Two beautiful initations with H. and with W. Got no words to describe this. Still integrating yesterday's realisations, today I suddenly felt why it is that I love skiing and snowboarding so much. It is a natural way to be here right now. Deliberate, eager even, and all the time. Because if you don't... :-) Quite interesting.
• Identity of the day: help, lacking data, I'm confused. Christine kindly shifts me out of it at the time that I can still really feel how the identity feels. Most helpful.

2008-03-07

• The students are source now and doing well, all moving into the reality creation list. H. amazes me spontaneously writing a beautiful list of personal primaries. When she sets to create them, hardly any secondaries. Experienced a very interesting bit of conciousness: for a while my hands did not feel as belonging to me. Got them back.
• Doing a Source List I realised that the reactive thinking, imagining, going into my head is a sort of non-deliberate flight. Not having to be here right now. I could feel that this is a small step towards the interesting experience of today. I'll have to deliberately decide to stay here more.

2008-03-06

• A beautiful day. At lunch time, W. comments not having felt so good as this since childhood times. At the end of the day, the perspective stretches into: I cannot remember ever feeling as good as this. Transformations like these are what really inspires me to share this with others.
• Identity of the day: I know it all, I am a superior master. Ugh. Quite NOT very helpful. The finding, feeling and discreating of this one already helped me a lot.

2008-03-05

• Today is Source List day. Managed to appreciate and so handle most of the student's creations quite effectively, without having to resort to active corrections.
• Anita helps me with a suddenly appeared planning game after she advises me to keep breathing. Most helpful in feeling how this identity of the day actually feels: I'm totally lost and confused, HELP. I'm not writing the other, darker side of that one here ;-)

2008-03-04

• The label exercises lead to great (and not so great :-) discoveries all over the place. I hit some creations today, but manage to not drown in them, regain control and relabel them as lessons. Sooner or later.
• Marieke is quite busy today but still has the alertness to be a kind mirror for me to help me discover my identity of the day: Things are not going the way I expected and I will NOT accept this. A very nice one.

2008-03-03

• Up early. Both W. and H. immediately get their feel-its. I fluke a chance that Marieke gives me that could have made it even easier for H. No harm done, valuable lesson learned.
• Checking my Novell email I'm quite pleased to find that the layout CWS has been nominated for inclusion, just in time for OOo 3.0. Yay! This is the first time I bring my laptop to an Avatar course/delivery. I worked hard to get the layout CWS integrated for OOo 3.0 and I promised to attend to any small issues that might pop up. Luckily, none did.

2008-03-02

• Up early. Second day of the workshop. Nice results on the finding transparent beliefs excercise. Paul recognises and describes my coaching experience with W. on this as a delayed explosion of a depth-missile. The realisation does not come with a big shift or the instant freeing up of attention. Just a small Hmm, and a rye smile. Some talk. 30sec later: BOOM, the big realisation. Emotions. Relief. Just beautiful.
• It's fun watching Eric help W. to go just a notch deeper towards the ownership belief in the signals of self-deception excercise. Impressive results with the workshop.

2008-03-01

• Up early. The first exciting day, welcoming the students of our table: C., F., H., R. and W.
• A very nice first day of the Resurfacing workshop. We work through the sections Orientation and Attention. The students are doing quite well, all having good results on the releasing fixed attention exercise. Sharing perspectives gets even more fluent and interesting towards the end of the day.

2008-02-29

• Up at 5.45. Nice and quite drive to the Avatar alignment meeting. North and middle part of the Netherlands have spring holydays: hardly any traffic jams.
• I'm sharing a table with Conny, Eric, Kees and Marieke. It turns out that Kees is also my roommate and he was guy who left a voicemail that I returned by leaving a voicemail.

2008-02-25

• Find that I'm on a dynamic IP address and that port 25 (OUT) is redirected. Request static IP. Manual blog refresh.

2008-02-24

• A pleasant surprise: found a mail titled free LilyPond advertising in lilypond-user. The title suggests spam, but it actually links to a very nice evaluation of Finale and LilyPond in three installments, by Andrew Hawryluk.

2008-02-23

• Update zypper build from SVN. Found that searching is now also fast, and searching descriptions also works (alas: you have to use search -d for that). These zyyp guys are working hard. Wondered why LilyPond is not found when searching for notation or sheet music, but zypper info says

  <!-- DT:Rich -->
  <p>GNU LilyPond is a music typesetter.
  </p>

  How is that for helpful? Time for another bug report :-)

2008-02-22

• Install SunStudio11 on SLED10. Found no compiler.
• Attempt to build OOo using SunStudio12 on SLED10, i386.
  • Find that included stlport causes makedepend to segfault
  • Manage to compile stlport, soltools, external, xml2cmp, sal, ignore Checking DLL ../unxlngi4.pro/lib/check_libuno_sal.so.3 ...: ERROR: ../unxlngi4.pro/lib/check_libuno_sal.so.3: undefined symbol: main
  • Give up in salhelper:

    sunstudio12.../ld: unable to find version dependency `UDK_3.0'
    sunstudio12.../ld: unable to find version dependency `SALHLP_1_0'
    sunstudio12.../ld: anonymous version tag cannot be combined with other version tags

  • Layout compiles fine.
• Mail Thorsten layout background bits.

2008-02-21

• Install SunStudio12, get it to compile (alas, no useful link).
• Find that I cannot reproduce Fridrich's weird compile errors, but that was an SunStudio12 snapshot and on Solaris.
• Fix all warnings.

2008-02-20

• Mail Joerg Skottke about resyncing and install sets.
• Junk VCLXProxy wrapper. Not moved into CWS, rather: http://lilypond.org/~janneke/vc/toolkit.git/.git
  • Do obvious thing for CancelButton, fail.
  • Bump head, tear hair.
  • Try less obvious things.
  • Complain to Michael.
  • Phone with Michael.
  • Redo obvious CancelButton thing. Look. Notice silent failure of VCLXButton::setLabel, setProperty.

  • Cleanup, for-all-buttons.

2008-02-19

• Get Joerg Skottke as QA contact for layout CWS.
• Buildbots:
  • Solaris/intel buildbot: successful on rebuild, but no install set http://buildbot.go-oo.org/buildbot/Solaris-Intel/builds/169
  • Windows 1 buildbot: weird temp file error on second build http://buildbot.go-oo.org/buildbot/Win-XP/builds/341 http://buildbot.go-oo.org/buildbot/Win-XP/builds/342
  • Install sets from: http://buildbot.go-oo.org/buildbot/Win-XP2/builds/27 http://buildbot.go-oo.org/buildbot/Ubuntu/builds/232
• Standalone test executable in CWS:
  • fix problem --inst DIR and inherited svapp/Init().
  • bump head against

    "Missing vcl resource. This indicates that files vital to localization are missing. "
    		  "You might have a corrupt installation."

    while running standalone TEST application from CWS. For the rest it works fine, except background is dark gray.
• Set ready for QA.
• Mail with Thorsten about dmake trouble in workben/layout, some build fixes thanks Thorsten!

2008-02-18

• Verify that all CWS patches are in anoncvs too.
• Fill in Release, owner, dates in layout CWS.
• Attempts looking junking VCLXProxy 1. remove it, have IconButton derive from vcl::Button_Base, build fixe, 2. revove IconButton altogether, work on vcl:CancelButton to display image.
• Trouble with resources: start a clean CWS build.
• Start buildbot builds http://buildbot.go-oo.org/buildbot/Win-XP/builds/341 http://buildbot.go-oo.org/buildbot/Solaris-Intel/builds/167

2008-02-17

• Eddy points me to an interesting analysis on an TDD-effectiveness study along with his report about Melbourne's flying chicken.

2008-02-16

• Found in lwn.net, but also everyone else copies the planetsuse headline that package management in 11.0 will be lightning fast. By this time, however, the blog item was no longer featured on planetsuse. Great. Found it at last in Google's page cache. Lightning fast means: even faster than yum. Hmm.
• Build zypper from SVN. Is it a coincidence that people who use SVN also use CMake? Anyway, dependency resolving time is indeed quite acceptable now, it's something like 200 times faster: comparable to apt. Still some work to do before 11.0, using zypper to search for a package e.g. still uses the old, slooow, mechanism.

2008-02-15

• Grand tralay cleanup, and limitations fix:
  • translate any and all _-prefixed attributes,
  • use own simple SearchL10NElements function,
  • rewrite InsertL10NElement function,
  • no set-content trickery and removal in merge mode.
• Tear hair out for a bit over cws-extract until I try cws-extract -m -d ':pserver:jcn@localhost:/cvs' layout Apparently the anonymous CVS server ':pserver:anoncvs@anoncvs.services.openoffice.org:/cvs' lags behind my the @localhost tunnel.

2008-02-14

• Commit new cws-layout patches: overnight build has succeeded and works.
• Use new cws-layout patches in src680/apply too, adapt the scp2.diff.
• Include layout-svtools into CWS.
• Add Fridrich's --disable-layout fix to CWS.
• Mail Stella Schulze about icons.
• Layout CWS cleanups:
  • Remove c&p in helper.cxx from vclxtoolkit.cxx,
  • Remove c&p in byteseq.cxx from xml_byteseq.cxx.

2008-02-13

• Get vanilla OpenOffice2 + CWS to install and run.
• Add images to CWS.
• Add layout docs and src2xml to CWS.
• Prepare new layout patches, extracted from CWS, ie, without top-level layout module.
• Start buildbot build to test CWS.

2008-02-12

• Use proper annotation of functions/classes to get layout code to link without removing -fvisibility hack.
• Checkout vanilla OpenOffice2 as a better cws development environment.
• Add config_office, solenv, scp2, svx, sw to CWS. Patch for layout code rearrangement. Start build.
• Get layout code to run in CWS build: standalone toolkit/workben/layout/test as well as the integrated dialogs. Used a partly unpatched ooo-build m245 as a base. This is a bit tricky, as running the test program needs an updated full installation (UNO-registration-wise).

2008-02-11

• Make CWS build, commit.
• Really apply kendy/Fridrich's parallel layout build fix.
• Remove tralay dependencies on sal and vcl.
• Move transex3 into CWS.

2008-02-08

• Write apply.pl.in hack adding or (||)-operator to support OOo-3.0, junk it after discussion with mmeeks and _rene_.
• Support OOo-3.0 in ooo-build through --with-tag=src680-m245, open patches/dev300/apply.
• Move layout CWS to SRC680-m245.
• Add offapi to layout CWS.
• Move all source files from layout/ into cws, except for src2xml converter.

2008-02-07

• Tralay:
  • handle top node translation,
  • some cleanups.
• Integrate compile time tralay run into layout/.
• Have layout parser strip leading _ from attributes as to allow displaying untranslated .xml templates too.
• Add patch to ooo-build.
• Create cws_src680_layout.

2008-02-06

• Tralay:
  • output options,
  • multiple languages: write to read-right tree: de/zoom.xml, nl/zoom.xml,
  • support fallback to en-US,
  • make helpex changes even less intrusive.

2008-02-05

• Get tralay to translate attributes (_label, _text, _title) instead of content.
• Implement extraction of translatable strings (create sdf) from attributes (_label, _text, _title).
• Team meeting.
• Remove code duplication:

  LayoutXMLFile

  allow inheritance & override XMLFile,

  TranslateLayout

  make & use static HelpMerge:CreateSDF & FillInFallbacks.

2008-02-04

• Try Ivo's xmlparse patch. This works, so I'm really grateful to Ivo for setting me up onto tralay.
• Look into layout usage of element texts vs. attributes label/text properties. Using Ivo's approach a new tralay tool would hardly be necessary, but it makes for really ugly .xml descriptions. Phone with Michael, decide to put a bit more effort into a translation tool to keep our .xml nice and clean.
• Start new tralay tool, reusing bits from helpex, look at translating (merge mode) first. Reads .xml and .sdf/ files, writes untranslated .xml.

2008-02-01

• Bugs.
• Send mail to Ivo Hinkelmann about src2xml conversion with transex3 questions.

2008-01-31

• Dentist.
• Bugs (slow start, until late).

2008-01-30

• Do iExpenses until it breaks, report bug.
• Fix patch/compile problems, test layout with m3.
• Find that I missed CNS training in week 2, catch up.
• Book flight.

2008-01-29

• Switch internet provider, now on glass fiber :-)
• Look into transex tools, no code yet.
• Team meeting.

2008-01-28

• First working day. Go through some mail, start a fresh build and test the new fiber glass network. It works, but my smc barricade WBR14G-2 dies with wlan speeds >= 10MB. Got a linksys WRT150N.
• Found that my simple changelog-find-file patch spurred a whole thread (28 messages and counting). Wondering whether that's good or bad. More viewpoints are always good, but there are so many sides to such a small thing. It reminds me of my responsibility about what subject to bring up. I could also have brought up the subject of embedding Emacs, think: evolution, firefox edit fields, eclipse, bash?
• Got a friendly email from Benjamin Weber kindly tapping me with a clue-bat. His webpinpackage search service is now installed on a decent server. It is fast and it worked flawlessly for me: it gives prompt results when searching for gtk.h now. Benjamin also pointed me at the webpin command line client. The easiest way to find out how to install it is using webpin :-)

2008-01-27

• Return from my Avatar® training. Wow.

  Thirteen days. The day starts with Tai chi. After breakfast a inspiring talk from Harry and then on to your morning program: simple but awesome and mindbogglingly clever experiential exercises that eventually trick your mind into giving up information that is transparent to you. Such information may be quite confronting--which is why the mind tries so hard to keep it secret--but very helpful once discovered.

  The rest of the day is filled with intriguing exercises that let you expecience new aspects of shifting viewpoints, attention, desire/resistance, identity, etc. Some lead to profound insights or disclose repeating patterns in your life. Others help you lay out the blueprint of your mind and change it. Or shift your mode of conciousness, or make you experience unbelievable connection with and compassion for others.

  A truely fantastic journey in conciousness, an empowering, humbling, mind broadening and joyful experience.

2008-01-10

• Realize that as an example language, German might be a better choice than Dutch, so add German translations too. Interestingly, the default buttons do not translate [to German]. Probably that's a local build problem as default buttons in old style dialogs are also in English. Leave this (as well as the simple teaser bug :-) to the fresh rebuild after my inspiring vacation.

• Must redo my expenses request. Apparently, I tried to file expenses too early: before I finished my eLearning on it. If you try to do that and go to the expenses reporting page, there is no link to do that (not even greyed-out or so), there is only iProcurement. Great. Finished my expenses eLearning, now wait two days for the iExpenses link to come up.
• Look into read-right translation of images. After a few fruitless attempts to reuse that code to read layout xml files, add another simplistic implementation to layout. Ugh. To get things going, also add manual translations of the zoom and wordcount dialogs and intergrate it all into OOo.

2008-01-09

• The layout dialog patches are now enabled by default in ooo-build, so go try them out! Configure using --enable-layout, ooinstall and open the zoom dialog. Especially if you are using a Sun compiler :-)
• Fridrich found that zoom.cxx does not compile with Sun's compiler due to some problems with cpp macro trickery.

  ( ( layout :: layout :: RadioButton * ) pCaller ) ; }
  	      long SvxZoomDialog :: UserHdl ( layout :: RadioButton * pBtn )

  It seems as if some macros are expanded twice, or that #undef does not work. Or quite possibly someone is trying to be smart, as we include layout-pre.hxx twice. We may have to remove the multiple inclusion shielding from that file and add another set of include headers layout-pre-header.hxx and layout-pre-source.hxx which simply include layout-pre.hxx again.
• Briefly look into installing a Sun compiler, but apparently it is not open source, and it seems to come only bundled with an IDE. Go figure.

2008-01-08

• Jakub has made a two beautiful sets of buttons. Now add them to the layout code.
  Too bad that the OOo toolbar also has an help button image with a life buoy.

2008-01-07

• Start another debugging round into translations, but start from the zoom.cxx end instead of our layout code. Michael calls with helpful hints and talk directions on layout i18n. Although it's interesting that some strings already translate, they shouldn't. We'll fix that and translate the xml layout descriptions at compile time.
• Find translation voodoo. Of course the magic is just plain code and it is right there in the Zoom dialog constructor. It is just the one place I just had not looked yet.

  	aOptimalBtn.SetText( String( SVX_RESSTR ( RID_SVX_ZOOM_OPTIMAL  ) ) );
  	      aPageWidthBtn.SetText( String( SVX_RESSTR ( RID_SVX_ZOOM_PAGEWIDTH ) ) );
  	      aWholePageBtn.SetText( String( SVX_RESSTR ( RID_SVX_ZOOM_WHOLEPAGE  ) ) );

  Duh. To be fair, this code should not have been there. I only debugged as far as initializing all buttons. After removing these lines this is what native OOo zoom box looks like. The texts that were translated are now missing. These prove to be introduced by the zoom-combobox.diff patch, which also removes them from the zoom.src file.

2008-01-05

• Michael was kind enough to hint me that I goofed when I gueimplicated that Yast2 is putting ugly caching bits in /etc/sysconfig instead of parsing /etc/* files. Fixed that. Mental note: try harder not to make judgments. If you found you did anyway, don't make them public. If you still feel an urge to do so, try to base them on more than guesswork.

2008-01-04

• Great. The rebuild fixed most localization issues, I now have a fully localized Dutch and German (en-US, nl, de) install. The cancel-button is translated (nl=Annuleren). The zoom dialog however has still three untranslated strings. The wordcount dialog is not translated at all.
  Oddly, they are exactly the strings that I added in my jcn.sdf. Now, do these still not translate, or did I break it? Not good.

  svx	source\dialog\zoom.src	0	radiobutton	RID_SVXDLG_ZOOM\
  	      BTN_USER	42	en-US	~Variable
  	      svx	source\dialog\zoom.src	0	radiobutton	RID_SVXDLG_ZOOM\
  	      BTN_USER	42	nl	~Variabelle
  	      svx	source\dialog\zoom.src	0	fixedline	RID_SVXDLG_ZOOM\
  	      FL_ZOOM		92	en-US	Zoom factor
  	      svx	source\dialog\zoom.src	0	fixedline	RID_SVXDLG_ZOOM\
  	      FL_ZOOM		92	nl	Zoomfaktor
  	      svx	source\dialog\zoom.src	0	modaldialog	RID_SVXDLG_ZOOM\
  	      SID_ATTR_ZOOM	160	en-US	Zoom
  	      svx	source\dialog\zoom.src	0	modaldialog	RID_SVXDLG_ZOOM\
  	      SID_ATTR_ZOOM	160	nl	Schaal

• Experimenting with zoom.xml learns that the ID controls if the Widget's text is translated. Changing the label has no effect at all. The odd thing is that the ID is removed by the xml parser and *seems* not to be used except as an index in a list.
• Remove my jcn.sdf and found the unxlngx6.pro/srs/*.srs intermediate files. So that's probably why removing the .res and partial rebuilds did not work.

2008-01-03

• Finally a partial localization success: some strings of zoom are translated. It turns out that these translations happen live in some other dialog's definition: stbctrls.src. Not translated: title (nl=Zet zoomfactor), fixed-line: (nl=Zoomfactor), radiobutton (nl=Variabel), cancel-button (nl=Annuleren). Adding hardcoded translations to zoom.src or merging my own jcn.sdf hack does not help.
• Ah, downloading src680-m241-lang.tar.bz2 and removing all */bin/.res files helps. Now a lot of things are translated, such as the crash dialog, but not any extra zoom buttons and not the main menu. My purposely weird strings are in /resource/svx680nl.res. Let's do a full rebuild tonight.

• Prepare and submit a build fix for xmloff with kendy's help.

2008-01-02

• Dump my collected notes about openSUSE into a openSUSE HOWTO survival guide.
• Reading my survival guide again I realise that by far worst thing for me in openSUSE 10.3 has been zypper. No surprise, apt is one of the tools I like most in Debian. In fact, I tried to use Debian in the era of dselect. That almost drove me mad and I turned back to Red Hat 4.2 (or was it 5?). Still it is unfortunate: zypper is the thing a Debian addict needs most after installing a new system. Otoh, mortals may just use the control center and not even know they're using Yast2 instead of synaptic. Maybe I should look into fixing some of the zypper bugs myself.
• I cannot help wondering: what has changed since Mark Shuttleworth's grand Debconf5 talk (ogg, mpeg). How much work is still being duplicated or falls onto the floor between distributions? I really like to think it was a passionate vision and not empty PR. I wonder what the relationship is between the Yast2 control center and gnome-system-tools.

2007-12-31

• The IDE->SATA replacment convertor arrives, install big disk at last. Setup raid1 for /home, /suse and /ubuntu.
• SUSE's apache setup serves /srv/www/htdocs instead of the /var/www that I bluntly copied . Wow, I missed that. Blog should be up again.
• Read up on FHS-2.3. It seems that Debian is violating the fhs using the /usr/lib64 -> /usr/lib and SuSE is OK with their /usr/lib64 directory. Apparently, Debian's excuse is consistency. There is something to say for that too as I have still got two /usr/lib64 related bugs on openSUSE 10.3 to file.
• It seems that the regex to strip the version number from an rpm is fairly simple: strip anything after a dash twice

  sed -e 's/-[^-]\+-[^-]\+$//'

  It helps if you know that the full name for libgtksourceview-2_0-0-2.0.0-5 is libgtksourceview-2_0-0.

2007-12-30

• T. mentions that the printer does not work. Find that updating or removing and re-inistalling a package on SUSE blondly resets configuration settings. I cannot remember having seen a warning or anything. Strange. Locate tells me I have now a number of of .rpmsave, .YaST2save and .Yast2.save config lying around. Is rpm taking the easy way out on the hairy config files issue? Have to look into this.
• It seems impossible to write a regex to strip the version from an rpm package. Looking at

  lilypond-2.10.29-24

  one is tempted to think that something like

  sed -e 's/-[0-9][0-9.a-f]\+-[0-9]\+//'

  might work, like in Cygwin. Consider

  SuSEfirewall2-3.6_SVNr183-10
  	      gpg-pubkey-1abd1afb-450ef738
  	      hal-32bit-0.5.9_git20070831-13
  	      cdparanoia-IIIalpha9.8-626
  	      iso-codes-1.0a-49
  	      libgtksourceview-2_0-0-2.0.0-5
  	      libopenssl0_9_8-32bit-0.9.8e-45.5

  I gave up anyway. I found a way out, instead of rpm -qai I now use rpm-get-selections.

2007-12-29

• Remove alien sawfish packages because I figured that could mean upgrade trouble and replace them by ~/pkg using the .bashrc.pkg hack.
• Notice the `Distribution:' tag in `rpm -qai' and find still quite some packages from the i586 and i686 distribution, probably left-overs from the yum install mess. Run a broken misguided script to remove all foreign packages by reinstalling using zypper. It stops when zypper itself is uninstalled, along with major parts of GNOME. Bummer.
• Install gnome-desktop, but that does not install gdm, beagle, yast2-*,... Search for SUSE's GNOME desktop meta-package (equivalent of `ubuntu-desktop'), but cannot find it. Realise that zypper keeps a log and recover doing

  grep "rpm' '-e'" /var/log/zypper.log | tr -d "'" | sed 's/.* //' > removed

2007-12-28

• Update to latest m241, type make. Laptop arrives! Play with SLED 10 and keep myself from installing a proper a 64 bit version of 10.3. Another not-so-productive day.

2007-12-24

• Took the morning off and bought new beds for A. & B. Looking for a way not to lose touch with Ubuntu, I found how to run the openSUSE installation in KVM: using loopback, linear raid and fdisk voodoo. I didn't succeed in the fdisk voodoo: I used parted and set unit to Bytes to skip the 16384B mbr offset.
• Next, I needed a kvm-switch switching mechanism to switch between using Native or KVM config files. Some extra work needed here: I want the native host always to provide all services of peder.xs4all.nl, the KVM host obviously must not. And, there must be a better way of switching configurations.
• When it worked I found a stability problems with NFS so it's back to fulltime openSUSE for now.

2007-12-21

• Find typo, fixes build.
• Allow other buttons in <dialogbuttonhbox> too. As an example, add an xml mockup of openSUSE's 10.3 shut- down applet dialog, which has a quite a nice number of extra buttons, again formatted for the various platform options.

• Share more code with Box.
• Fix zoom.xml: custom-unit-text for m241.
• Prepare & commit new layout-dialogs patch.
• Install and read .xml files in/from BASE_INSTALL/share/layout (WAS: BASE/program).
• Prepare & commit new layout-dialogs patch.

2007-12-20

• Update to m241.
• More standard buttons for <dialog- buttonhbox>: Yes, No, Retry, Ignore, Reset, Apply.

• Some cleanups: default to border=0, padding=0.
• Prepare & commit new layout-dialogs patch.

2007-12-19

• Help Joerg and get awtfixes1 into Verified!
• Implement <dialogbuttonhbox>. As an example, add dialogbuttons.xml.
  Also use it in the wordcount
  and zoom dialogs. This already looks a whole lot better. Now for the proper icons...

• Finish pension paperwork.

2007-12-18

• At last, get back to working on layout code: Ok/cancel/help button ordering box: <dialogbuttonhbox>.
• Help Joerg Skottke to Windows and Ubuntu install sets.
• Team meeting.

2007-12-17

• Quickly look at plugging mysql instead of sqlite into zypper, seems to be more than a 10min hack.
• Set parent of Dialog. Fixes critical modality bug.
• Update ooo-build layout-dialogs patch.

2007-12-14

• Found that Solaris-sparc build was succesful, error uploading packages.
• File a couple of bugs

  #348669 - zypper should print progress to stderr  
  	      #348676 - zypper should list package descriptions/summaries  
  	      #348683 - Zypper [info] is amazingly slow  
  	      #348685 - zypper should list or install build dependencies for package  
  	      #348733 - zypper should have --download-only and --cache options
  	      #348738 - xorg.conf is unclear about editing, regenerating/overwriting,
                inserting [openchrome] driver  
  	      #348741 - metacity's maximize-vertically makes windows too tall

  and vote for some others. That should help get us up to speed as development platform :-)
• Try-out Ubuntu awtfixes1 install set, seems to works fine.
• Compile fix for layout with debugging.
• Look at modality bug. For old dialog, using NULL parent

  	      //  SfxModalDialog( pParent, SVX_RES( RID_SVXDLG_ZOOM ) ),
  	      SfxModalDialog( 0, SVX_RES( RID_SVXDLG_ZOOM ) ),

  does not make the dialog un-modal. Also, our Dialog::setParent () is empty. Will look into SetParent on Monday.
• JP fixes up my sawfish spec files and adds them to his build service home. Now I'm wondering what will happen when the system is updated. It's probably wiser to install missing software in ~/pkg instead of installing foreign RPMs.

2007-12-13

• Only slideshow didn't build, can't find offending patch to revert, so do slideshow/source/engine/shapes/viewmediashape.cxx

  #if 0
  	      mpMediaWindow->SetForwardKey( TRUE );
  	      #endif

• In meantime, write silly zypper-build-dep script.
• Layout works standalone and integrated on SUSE-m239+awtfixes1!
• Found /etc/sysconfig/displaymanager. Wow, also found

  /etc/sysconfig/displaymanager:DISPLAYMANAGER_XSERVER_TCP_PORT_6000_OPEN="yes"
  	      /etc/gdm/gdm_sysconfig.conf:DISPLAYMANAGER_XSERVER_TCP_PORT_6000_OPEN="yes"
  	      /etc/gdm/custom.conf:DisallowTCP=false

  some heavy caching going on...
• Distribute sawfish specs and run createrepo (src.rpm and x86_64).
• See after buildbots. Got two succesful builds Windows XP and Ubuntu. SUSE-10 is offline? Started Solaris-sparc one.
• Gather info on openchrome drive breakage. After adding some debug printing, I get

  (II) VIA(0): URG Accepting:
  	      CrtHDisplay (1680) * CrtVDisplay (1050) * VRefresh (59)
  	      * bitsPerPixel >> 3 (32, 4) = 423037184 > 74000000 (biosinfo: BandWidth)

  so it seems there is a problem with the biosinfo: disabling this check makes the driver for my P4M890 work.
• Learn about openSUSE's equivalent of Contents.<arch>.gz, which is quite handy as SUSE does not seem to have things like packages.debian.org or apt-file yet.

2007-12-12

• Update jsk about m239 resync, set awtfixes1 ready for QA.
• Ping ause about m239 untar failure, now m239 fails on XP in scp2 and bridgetest/cli. Apparently there's a problem on windows to build < m239 and >= m239 at the same time.
• Eventually get 1680x1050 up. Needed to comment-out the BIOSInfo->Bandwidth check. Added debug info and will check in Gutsy some time later. Then spend some time before realising that hw cursor is broken.
• Hunt down svn archives for, update spec files, install and build librep, rep-gtk, sawfish.
• Found that m239+awtfixes1 does not run, but does not look so nice

  (gdb) bt
  	      #0  0x00002b016bcdfcc9 in ?? () from ./libvcl680lx.so
  	      #1  0x00002b016bce5d2d in MetricFormatter::Reformat () from ./libvcl680lx.so
  	      #2  0x00002b016bce5e39 in MetricBox::ReformatAll () from ./libvcl680lx.so
  	      #3  0x00002b016c4d2b60 in ?? () from ./libsvt680lx.so

  but also found that default build [on SUSE now] is without -g?
• Do a new svn up and start clean rebuild to run overnight.

2007-12-11

• Build OOo m239 + awtfixes1 on openSUSE 10.3.
• Identify and report bug on yum system breakage (thanks Fridrich!).

  #347624 - yum installs i586 packages on x86_64 which easily results in serious breakage

• Get proprietary flash going with nswpluginwrapper and flashplugin.
• Team meeting.
• Start buildbot build on xp machine (Thanks, Jan!)
• And just before going to bed I see

  Build succeeded ...!
  	      touch stamp/build

  :-)

2007-12-10

• Search all the web for yum/rpm/yast way of typing `apt-get build-dep', install BuildRequires: packages from OpenOffice_org.spec using yum. At the y/n prompt blondly press y. Find my brand new SUSE installation broken after that.
• Spend time playing with rpm and yum removing arch-i586 packages (35 of them) that got installed with partly resolved dependencies. (Have to look into this; it seems that an arch-i586 package can somehow satisfy a x86_64 dependency, so yum installed a mix of x86_64 and i586 packages, with i586 libs (eg libXdm) depending on x86_64 (eg libidn) ones. This broke sshd, gdm, login etc).
• Discuss state of awtfixes1 and conflicts in resync with Michael.
• Make minor fix, test NOLABEL feature (true/false) using zoom dialog, debug compile fix in layout. Set awtfixes1 to Ready for QA.
• Resync awtfixes1 to m239.

2007-12-09

• Install openSUSE 10.3. I'm presented with an EULA that I must accept or otherwise the installation process is aborted. Weird. For a moment I wanted to stop right there. As far as I know, EULAs aren't even enforcable here. But what do I know, how great am I at selling free software? An EULA may piss off a couple of freedom hippies but might be essential for paying newcomers, giving them warm fuzzies. Who knows such things. I take a deep breath and continue. The rest of the install process is quite smooth and mature. It detects that it cannot make a bootable installation because of my lvm setup (there's no grub2 in SUSE yet) and accepts that with a big warning. Nice. Even cups and the eth0:0 alias can be configured from the installer. I did need the console to figure out that the reason I cannot enter the `:' is that I must enter `0' to get `eth0:0'.
• The installer looks a bit like Debian's installer, but the menu with installation steps is not clickable, making it less flexible. Like Debian, it is less targeted at (or ready for) newbies. You need to answer *a lot* more than Ubuntu's eight or so simple questions. You do end up with a fairly complete configured machine, though.

2007-12-08

• Test grub2 lvm setup, copy results, prepare and send bug report to grub-devel. Got no quick answer, though.

2007-12-07

• It turns out that my SATA to IDE converter that I got with the hard disk is the wrong one, so send it back. I stop my raid in favour of making a mirror on the new disk, so that I can still install SUSE this weekend. Also, I must have messed-up my laptop request, because it has been over a month.
• Learn that I can do insmod normal to start Grub2's menu from the rescue mode, or else see the errors that prevented it from starting. Will try that to send a rfh/bug report.

2007-12-06

• In preparation of installing openSUSE, looked into backup options. Had a go with SimpleBackup, which looks sort of nice but does not seem to go anywhere fast soon. Make simple tarball backups while reading about backup tools. It seems like all backup software besides tar and dump is for large pc farms (Amanda, Bacula), or it is collection of perl scripts (BackupPc, hpdup16 → hpdup2 → rdup, rsnapshot ) or it is in beta while development is discontinued (SimpleBackup, Dirvish). Not much use but possibly interesting to watch are alfa and development pyBackPack duplicity and rdiff-backup. There's a helpful though stale comparison sheet on BoxBackup's site.
• During dinner my hard disk arrives. Spend until late playing with Grub2 and lvm. Grub2 looks quite nice but I cannot get it to recognise my lvm devices at boot time. The 20070604 version will boot with the lvm module loaded, but fails to see any lvm device. The 20071101 version boots into rescue mode as soon as the lvm module is added to core.img. If I insmod lvm by hand or in the grub.cfg, then it works fine, but that still means that grub.cfg cannot live on lvm. Decide to put a grub directory with grub.cfg on a non-lvm partition for now.

2007-12-05

• Now that I have a funtional build tree that's up to date I can test the integration of the dialogs with OOo. No surprises, the clea build has also brought the default icons back. Jakub Steiner will be getting or creating us a set of nice GNOME icons.
• Break early for Sinterklaas party. A.&B. get spoiled with presents, and are too excited to fall asleep on the drive home.

2007-12-04

• Up early. Spend more than my day building the cws (child work space) awtfixes1, the patches that underlie the layout code. Finally, when A.&B. are in bed, I find that the build finished cleanly and the layout test works.
• In the mean time I make some preparations for installing openSUSE and order a new hard disk. However much I love Ubuntu (polished, stable, bleeding-edge Debian unstable without weekly/daily update drill/nightmare: developer's galore AFAIAC, *and* ready for my Mom and Dad out of the box without more than an evening's tweaking), I will now get to use and love openSUSE too! Playing with kvm is more hassle than I thought (there's still the X buffer overwrite problem in 2.6.22 and I don't feel like baking a .23).

2007-12-03

• Joyful news: our brothers in arms at SUN are reading my blog and like our layout work too! It turns out that the OOo's static resource style layouting is something people would like to see replaced. SUN may be looking more into new dialogs, but that's all for the best. There's already a feature request from Christian Lippka, which is about OK/Cancel/Help button ordering.
• Each platform--and OOo supports a few--has its own guidelines (or rune at least) for alert button ordering and naming. In our new layout code everything should look like the platform expects, so that OOo blends in with the rest of the UI instead of looking like a monster. Mozilla has come a far way in that area, esp. if you look at Firefox 3.

• Here is what OOo currently does on GNOME. If you are used to GNOME it feels a bit funny.

• That is because our hig guidelines say that alert buttons should look like this.

• Here are the buttons of a typical GNOME save alert.

• We will probably introduce a smart hbox that will rearrange the buttons depending on the platform the code runs on (or is compiled for). Something like

  <alerthbox>
  	      <okbutton     id="BTN_ZOOM_OK" default="true"/>
  	      <cancelbutton id="BTN_ZOOM_CANCEL"/>
  	      <helpbutton   id="BTN_ZOOM_HELP"/>
  	      </alerthbox>

  should then produce a native ordering for different platforms.
• Along with this feature request of ordering we may also want to customise the label on the OK/Cancel/Yes/No buttons for different platforms. For GNOME (and probably MacOS X) we will then not use the default OK/Yes/Cancel/No texts, but a custom label that the dialog specifies. Of course, all that text needs to get _("localisation") tags. Possibly something like

  <okbutton id="BTN_SAVE_OK" default="true"/> label="_(~Save changes)">

  the localisation marker will probably be different.
• On Windows XP, the default OK/Yes and Cancel/No texts can be used if that is what XP users expect (Ja=Yes, Nee=No, Annuleren=Cancel in Dutch). How do I run LANG= msword.exe?

2007-11-28

• The build was ready, so I can test the zoom dialog. It starts fine, but where do these retro icons come from? I decide to remove debug printing from the layout code as most of it is tracing. I notice that the code could also do with some whitespace cleanups before it goes into ooo-build, so I script some to fix the worst offenders. There are also lots of unnecessary includes and remove two catch-all globals.hxx headers, prune most of them. There, ready to make my first commit to ooo-build!

• After another round of svn update, make (ie: patch.unapply [fix-fix], patch.apply[fix-fix]), prepare final [Layout] and [LayoutDialog] patches. As I'm ready to commit I find that the protocol svn:// that I changed all the http:// entries to also means read only. What are these subversion people thinking? I realise that I must pick-up A.&B., so I'll commit after dinner.
• Dinner. A.&B. to bed. Use svn+ssh:// (yes, I guessed the other way around, does not work). [My first!] Commit. Wait. No loud explosions. Another very nice day.

2007-11-27

• Today I'm going to port my layout patches to m237, so lots of GIT juggling for me. Which means git format-patch, git am, git rebase, git rebase -i, with a touch of svn update, make patch.apply, make patch.unapply. I now only have the modules that I have changes in in GIT. It would be safer and nicer to just put everything in GIT, then I could simply run a script with

  git branch -f svn-up patch.apply
  	      git checkout svn-up
  	      make patch.unapply
  	      svn update  # or, git-svn rebase :-))
  	      make patch.apply
  	      git commit -a -m 'Svn update.'
  	      git checkout master
  	      git rebase svn-up

  and everything should be fine (patch.apply and patch.unapply fail sometimes and rebase may find a conflict, of course). But I can't do that because my patches live on top of ooo-build's patches, so I just have to hope for the best. If something goes wrong, there's always rm -rf and git format-patch, git am :-)
• The GIT juggling and re-builds leave me a lot of time to hack on Kohei's src2xml converter. I add the postprocessing and then some, so that the zoom dialog runs straight from the converter. It turns out that the hbox and vboxes are not specified in the original .SRC, so it seems we'll really want to massage the xml manually.

2007-11-26

• Preparing an m237 build tree. This is a mostly waisted day, from a coding point of view. I did manage to hack the scp2 snippet and makefile to install the zoom and wordcount xml files. Other than that, I learned not to use --with-tag or rather, the universe taught me once again not to use !./conf too casually.

2007-11-23

• Spent most of my day preparing [LayoutDialog] patches to go into ooo-build. There's one interesting thing to show today, I put the OK and Help buttons of the wordcount dialog inside an <hbox homogenous="true">. Eventually we will need a special hbox that is platform-aware for Cancel/Help/OK buttons, that will automagically order the buttons according to the platform's rune.

2007-11-21

• Found that my new sandbox build has not completed. Check on irc if m9 is the way to go. Fridrich must run but _rene_ is helpful. m9 is using the -fvisibility-inlines-hidden feature and system stl by default without checking. My system stl is broken wrt hidden inlines.
• Devise a couple more macro hacks to make zoom and wordcount diffs real even smaller and hack up a script to distribute these files back into the OOo tree.

  index caeddf2..b801ada 100644
  	      --- a/sw/source/ui/dialog/wordcountdialog.cxx
  	      +++ b/sw/source/ui/dialog/wordcountdialog.cxx
  	      @@ -50,6 +50,7 @@
  	      #endif
  	      
  	      #include <dialog.hrc>
  	      +#include <layout/layout-pre.hxx>
  	      #include <wordcountdialog.hrc>
  	      
  	      /*-- 06.04.2004 16:05:55---------------------------------------------------
  	    

  index 2030487..1c96157 100644
  	      --- a/sw/source/ui/inc/wordcountdialog.hxx
  	      +++ b/sw/source/ui/inc/wordcountdialog.hxx
  	      @@ -43,6 +43,8 @@
  	      #ifndef _SV_BUTTON_HXX
  	      #include <vcl/button.hxx>
  	      #endif
  	      +#include <layout/layout.hxx>
  	      +#include <layout/layout-pre.hxx>
  	      struct SwDocStat;
  	      class SwWordCountDialog : public SfxModalDialog
  	      {
  	      @@ -70,4 +72,6 @@ public:
  	      void    SetValues(const SwDocStat& rCurrent, const SwDocStat& rDoc);
  	      };
  	      
  	      +#include <layout/layout-post.hxx>
  	      +
  	      #endif

  Now that's more like it.
• Although my m9 sandbox is not built yet, go on and prepare the [LayoutDialog] patches. Then cd sends small list of things that need to be fixed for the [Layout] client workspace. I fix them but do not rush to commit. Go in plenty time to collect A.&B. from Kersentuin. Work late again: after dinner, commit fixes and keep m9 build going. Wondering what is faster: removing all unxlngx6.pro directories, or removing files complaining like

  undefined reference to `std::allocator<char>::allocator(std::allocator<char> const&)'
  	      undefined reference to `std::allocator<char>::~allocator()'
  	      hidden symbol `std::allocator<char>::allocator()' isn't defined

  and keep kicking it?

2007-11-20

• Finish my layout status talk. Michael suggest that I prepare ready-to-use [LayoutDialog] patches soon. I thought to postpone that to do some more fixing, but you can never release too soon. I start rightaway setting up a clean tree, alongside the half patched, half pristine, half pristine+compile+[Layout] patched one that I'm working in.
• The diffs that I show in my talk are still too big. hub and jonp suggest typdef and macro trickery, Michael elaborates: use layout-pre.hxx, layout-post.hxx header. After the team meeting I set to some evil macro hacking. The diffs are now barely more than including layout headers. Just get it done, must hurry to pick up A.&B. After dinner, I start setting up a fresh oog680-m9 sandbox.

2007-11-16

• Start the day with a fun task: getting the percent (%) sign to work in the zoom dialog. I decide to just go looking in the implementation of MetricField. This tells me that the Custom Unit property is not set. I knew this, it is commented-out in the zoom.xml file because it does not work. Enabling it shows that we do not grok properties of enumerated type, such as Custom Unit. Interestingly, enums are 16bit in Uno. Anyway, an easy task with a nice result.

• Write an scp2 snippet to register our Layout module the proper way. I must be overlooking something, though. The ./layout directory is included in the dmake build process, but the layout library does not get installed and is not registered.

2007-11-15

• Spend most of the day bug hunting for the second invocation bug. Learn a bit more about component registration and references. Mail with Michael who is on leave these days (shame). Finally find a workaround: just disable the check and just re-initialise an already initialised LayoutRoot.
• Inspired by the word count dialog tweaks, add a couple of tweaks for the zoom dialog too. Wider borders, add some space between the `Variable' label and spin button box. An experimental addition is an extra indentation for the list of radio buttons, but this hack adds another pair of hbox and vbox, which makes for ugly resizing.

2007-11-14

• On to porting a new dialog today: the word count dialog. That does not seem too difficult, right? Also, I `discover' Kohei's converter which generates the first default layout, almost ready to use.
  As it is a bit tight, I add some spacing. Now get the code ready and plug it into OOo.
  But wait, the zeroes are not really aligned, are they. The top zero is a few pixels to the right. When the dialog is widened, the upper zero `jumps' to the left but now it is a few pixels too far to the left. Funny bug.

2007-11-13

• The crash with the second use of the zoom dialog is caused by the radio buttons. There is a hack to group them: when a button is created it is added to the group list... but it is never removed. Wow, the zoom dialog is now almost perfect. Next thing is to register the Layout module the proper way: there is a tool called regcomp to do that. After I use regcomp and remove my custom Layout registration code, I find out that the regcomp registration process triggers another second-use-bug in the Zoom dialog.
• I have a look into the %-sign that is not displayed in our zoom box. Not sure how this works. Adding FUNIT_PERCENT does not fix it. We do set a property custom-unit-text=" %". Hmm. Other than that and the new crash that I found today, the new zoom dialog looks perfect! Just to remind you I've put up a picture of the classic OOo zoom dialog below. Now how does it do that percent (%) sign? ;-)
  

• Team meeting. Collect A.&B. from De Kersentuin.

2007-11-12

• Up early. Find that the zoom dialog is not functional because initialisation and selection code is switched off. After switching it on, debugging tells me that it always returns Cancel. Fix that too, merging Cancel/Help/Ok all into one EndDialog function. Very nice. We now have a fully functional Zoom Dialog implementation, although it still craches the second time it's used :-)

2007-11-09

• Up early. Add registration for Layout module in several places until soffice.bin does not start up anymore. Start a build at toplevel... Fun.
• See that images and text get messed up in my blog's rss feed. Add style-sheet to the rss feed and find firefox ignores it. Hack until it works. Meanwhile try several more things to hook up Layout module registration. Go for main () but not much happens. Deliver does not seem to work and cp does not produce a working soffice.bin? Finally remember about ooinstall. Another core dump. Try several things then stop trying and think about the core dump.
• T. arrives home with a sick A.&. Spend some time with her, back to work. After a bit T. brings her and A.& sits very quietly on my lap. Start reading the standalone test code, remove `Weak' part from bootstap code and make a full implementation. Yes, the Layout zoom dialog pops-up from the View/Zoom menu, and the Help button seems to work. Just in time for dinner. Make mental note: stop trying.
• Have a strange dinner in stages, bring A.& to bed early. Finish up and commit code.

2007-11-08

• I have been using Evolution 2.12 for a couple of weeks now as my main MUA and found it quite usable. There's one thing that keeps annoying me: focus follow mouse is broken in Evolution's subwindows. How do people cope with this?
• Make new layout zoom dialog code fit into OpenOffice svx tree with minimal changes. Get OpenOffice to build with both the old and new layout flavour of the zoom dialog. No luck in running new dialog though, registration of the Layout module is missing. Nice quiet evening, manage to go to bed early.

2007-11-07

• Up late letting T. handle A.&B. while watching my bowels come to terms with the world. To work.
• Hunted down the extra pixel in the Variable radiobutton's left margin. This fixes all known visual bugs in the zoom dialog! Added Help and OK interface similar to Cancel.

• Got a reply on my change-log-find-file patch: learned about ffap.el. Appended

  (require 'ffap)
  	      (ffap-bindings)

  to my ~/.emacs.el which makes C-x C-f behave in the way you would expect.
• Off to fix the Variable radiobutton's behaviour but found that the Variable user radiobutton actually activates the metric field. This may well have been related to the vtable problem resulting in the disfunctional OK button. Well well, if bugs go fix themselves as quickly as this :-))
  Found new visual bug rightaway though: see what happens if the metric field value is increased to 10. You can just see the edge of the 10's zero at the right. So, the metric field is too narrow by default, or allows too big values, and/or does not resize (correctly) when the contents get bigger. Leaving that for now.

• Hack rss-ize(.pl, yuck :-) script that Michael sent me, add link to page. To bed early, quite a bit later than planned.

2007-11-06

• Played with zoom.xml and ultimately found that extra pixel is in the hbox. Looked into Noel's coredump. Team meeting. Collect A.&B. from De Kersentuin. The virus that currently floors everyone in Holland for a day but seemed to miss me finally got to me too.

2007-11-05

• Found I left some of my work at Michael's, so start my working day with HR stuff, sign forms, filing travelling expenses.
• Thought about starting a blog (again) and found an email from Michael wondering about my blogging activities. Ah, such coincidence! Looked at and dismissed experimental blogger.xs4all.nl and blogspot.com for obvious reasons. Started simple HTML blog, find way to rss if the need arises.
• Merge layout efforts and get latest OpenOffice.org layout code up and running in my child workspace. Spend evening tweaking and filling blog, to bed late.

2007-11-04

• Have fun with A.&B. while T. rests. Got a mail saying that the mutopia project has reached a 1000 pieces of sheet music today. Eat at Peter and Marleen's for Daaf's birthday.
• Rediff change-log-find-file feature against add-log.el CVS, submit patch to emacs-devel. To bed early.

2007-11-03

• Up at 4.55am. Skip coffee when finding that taxi ordered for 5.15 has already arrived. Leave hastily written goodbye note to Michael and family on doormat :-) What a grand visit this has been: fun, inspiring, warm and crammed with information. Off to Stansted. Find it crowded by lots of barely awake people while the rest of the country seemed to be sleeping. Most of the personnel seems friendly and quite awake. Weird places, airports.
• Got an SMS from T. saying they would pick me up as they did. Enjoy A.&B. enjoying watching the airplanes getting ready for take-off. Have coffee with shared cake and brownies. Drive home, got a present: De Koe ging over de Berg, one of my childhood favourite golden books. Treats with childhood favourites are in the air... T. shows me a CD she was compelled to buy and plays a song to me: Halleluja, halleluja... Just like we sang at the gathering. Interesting world, this is.
• Work in the garden with the family, fix the pavement in the corner under the car port. Go shopping with the family and buy all sorts of useful things, such as a new doorbell that we will hopefully actually hear. Treat myself on a fresh pair of clogs, ditch my favourite Swedisch pair after ~20y of duty. Not bad.

2007-11-02

• Couple of small tweaks to zoom.xml. Almost there. The ugly margin is gone and is not expanded any more when the dialog is stretched, but we are still losing one pixel somewhere.

• Duplictate child workspace setup on peder (my home pc) now that's Michael here to assist. Collect and upload my work from Michael's pc. Break early for trip into Cambridge.
• Meet guys from Collabora in their new office right opposite King's college. Visit the amazing King's Chapel while Michael makes yet another confcall in Collabora's office. Have a nice walk around some of the colleges, visit the round church and noted some places of scientific history; announcement of DNA, discovery of the electron.
• H. demonstrates PlanetPenguin racer and simple things from GCompris. Just perfect for A.&B. Watch the Czeck Mole (one of my childhood favourites from `Die Sendung mit der Maus'). A home-made pizza dinner, watch talk with Julia and Michael who has to say goodbye, has another late night conf call.

2007-11-01

• Up early, enjoyed the extra hour of morning light, special breakfast. See and hear even more lots of interesting stuff. Fix height of FixedLine in the Right Way(TM). Get accounts on openoffice.org and gnome.org.
• Fix width of SpinField button. It's especially noticable at bigger font sizes that the width of the spinfield button is not calculated correctly. It turns-out not to be really calculated at all :-)

• Order laptop. Setup child workspace cws_awtfixes1 and add MetricField patch files in issue i#83222#. Work late and right off to bed. Another fun day.

2007-10-31

• Up early and just caught A.&B. awake to say goodbye. Bus to the airport. Amazed at how exactly on-schedule they are, waiting 20s. or so at a bus stop here and there. While queueing for 30mins and re-reading all the warnings about the new fluid restrictions I realise that my toothpaste needs to be in a separate plastic bag. My toothpaste might be part of a bomb, go figure. Completely forgot about my swiss army knife in my hand luggage though (no obvious warnings on knifes, guns, swords etc. anywhere) and lost it.
• Found Michael just as he arrives, nice ride to Newmarket. Warm welcome, and get to work. Setup an account and mostly cripple it with my Ubuntu gconf settings, fight a bit with my `too bleeding edge' Emacs settings, commenting everything out until defvaralias error goes away, happily assisted by Michael: why not comment that out too :-)
• Make layout, vcl and toolkit patches digestable and talk/ask through them with Michael. Have fun for a while with the address of *this is in an (single) inheritance tree where the vtable gets introduced somewhere in a derived class. Revert backward PushButton dimension merger that I didn't understand :-). See and hear lots of interesting stuff, got served a lovely dinner by J. and enjoyed a very special evening with friends.

2007-10-30

• Add default click handler to Button, fix bug: Help and Cancel buttons do nothing.
• Another small but nice improvement: consider text height of a FixedLine that has text.

• Debugging into WindowImpl's weird understanding of mpWindow address. I must have broken this, but where? Import Kohei's converter from ooo-build svn. Team Meeting. Off early to say goodbye to Xin Li@SES (no homepage, Chris? ;-), found that Joeri is in and really wants to return my DVD that he only watched part of and all others happen to be away. Makes me wonder ;-). Pick up A.&B. from De Kersentuin.
• Didn't forget to do the online check-in, pack my stuff including my now rather broken laptop for the trip tomorrow.

2007-10-29

• Continue on bug 3: Help and Cancel buttons do nothing. Find out how to cancel the dialog and hook up the custom handler. Several more attempts at hooking up click handler.

2007-10-26

• Check Metric Field sizing, seems OK.
• Start with bug 3: Help and Cancel buttons do nothing. Add XDialog2, add default buttton click handler and attempt to hook it up.
• Read http://wiki.services.openoffice.org/wiki/Cpp_Coding_Standards (is that all, are naming and layout not explicitly specified?).
• Finish week 1 of New Hire checklist.

2007-10-25

• Start New Hire checklist. Mother arrives and A.&B. seem to already grasp the concept of T. and me working while they're at home. Amazing how quick kids are.
• Michael's off for a week but he left a very nice and soft introduction into the layout code in the form of a TODO list. The zoom dialog is a layout'ified copy of a real dialog in OpenOffice and apart from the odd little icon buttons it doesn't really seem to look so bad...
  ...until you resize the dialog and find what's missing... At the top there's a Zoom Factor----- widget getting squashed and the OK, Cancel and Help buttons have labels that are apparently ignored while calculating their width.
  Aah, much nicer, isn't it? Now the text label of the buttons is actually taken into account when calculating the size of the button. Fixed [my] first bug: button sizing. In: http://lilypond.org/~janneke/vc/ooo 6b55a198..9cb76668.

• Fix warnings in layout code: found that -Wall was added, also add -Wnon-no-virtual-destructor.

2007-10-24

• Finish COBE eLearing.
• Discuss example with Ricardo, verify fixes to layout. Dive into first layout bug: button size.
• Collect A.&B. from De Kersentuin.

2007-10-23

• Do most of COBE eLearning.
• Finish and report on layout excercise. Look at various levels of layout code.
• Team meeting. Collect A.&B. from De Kersentuin.

2007-10-22

• Add --janneke option to run SvxJannekeDialog explicity, makes simple button click callback work.
• Add Edit Box and hook it up, change text in three levels of indirection.
• Update GIT with latest tarball w. git from Ricardo. Lots of small conflicts. Amazed at finding that Emacs' ChangeLog-mode has not hotkey to visit file for the chance at point. Hack-up a simple patch for add-log.el, mental not to submit this later. Resolve conflicts and fix x86_64 build, new repo at http://lilypond.org/~janneke/vc/layout.git+tarball/.git

2007-10-19

• Report and discuss Feisty/Gutsy problems with calc. Make proper fixes to the layout build for x86_64. Add some scripts to build from Emacs. Setup Emacs indentation and camelCase (and fix it).
• Write simple standalone widget based on zoom but not yet quite what Ricardo suggested, I presume.

2007-10-18

• Finally book flight to visit Michael. Study ryanair.com's javascript until satisfied that it is broken with LANG=nl, use LANG=en, search invain for webmaster's email and get on with it. Try to explain A.&B. that I'm not there for them the whole day and that that's why mother is visiting us.
• Report ooo-build problems and fixes to _rene_. Checkout Ricardo's layout code, read layout.odp presentation from oocon. Build Michael's layout tree with Fridrich's help, stuck at not getting ../bin/test. Subscribe to mailing lists. Build first layout test with Ricardo's help.
• Quit working at a decent hour, spend time with T.

2007-10-17

• Get SuSE 10.3 desktop in KVM up and running, play wit YaST.
• Do My First Hack. Learn first bits about the build system (relocate, build alias etc). Get @novell.com imap up in Gnus. Found that xs4all.nl also offers imap nowadays (since last year). Register for eLearing.
• Phone with Michael. Collect A.&B. from De Kersentuin.
• Found that Gutsy was actually released during my holidays. That's what you get when you don't catch-up with your /. and lwn, I guess. Upgrade desktop to Gutsy, fixes libxul-dev/firefox/evolution breakage. Another late night.

2007-10-16

• First full day working from home! Must get OpenOffice to build :-). Decided to ignore my Debian i386 KVM and use my comfortable Feisty64 desktop as well as figuring a way to run SuSE i386.
• Find that apt-get build-dep openoffice.org lacks gperf and libxul-dev, libxul-dev is not installable on Feisty64, install xulrunner in ~ from source. Add --with-system-neon and add some kde and java symlinks in /usr/lib.
• Team meeting. Collect A.&B. from De Kersentuin.
• Work late: prepare suse1.03 rpmstrap script, as my kvm does not boot from OpenSuSE cd/dvd. Manually populate chroot to get yast running, boot-up using debian kernel/initrd.
• Checked ooo-build just before going to bed and found it built, yay!

2007-10-15

• First day at Novell! Although quite excited, spend an hour+ getting my plain OpenOffice.org build going before leaving for the Rotterdam office. Figured this way I would make the most of my travel time while beating the traffic jams.
• Had a very sunny, nice and quiet drive as after 10am the lorries are free to overtake again. Talk through all New Employee stuff; income tax, medical insurance, pension plan, spaarloon, etc. Had two pictures taken for badge, just in case. Lunch in the cantine with Ton, nice sunny drive back.
• Back home whine about troubles trying to build plain OpenOffice.org. Learn about ooo-build and --with-distro. Disable two patches that break build on Feisty64. Phone call with Michael.
• Leave early to eat at C's for H's birthday.
• Work late trying to make up for my lapse of reason the past few months not installing some 32bit version of SuSE. Knowing I'd be working for Novell, I did prepare a KVM running debian i386 unstable. Apparently that seemed my best bet at the time.

In case it's not painfully obvious and also in case it is: the reflections reflected here are my own, and don't reflect the views of Novell, or any other GNU.

---

For older activity log see change logs of GUB and (look way back) LilyPond

---

Other languages: Nederlands.

Jan Nieuwenhuizen (janneke@gnu.org)