I'll visit Barcelona with Elizabeth for four days. No laptop, no beersigning, no electronic stuff. Only sightseeing and enjoying time together without any geek or work thinking. That was desperately needed..:-) Back on Monday...

One of the things I do when I’m at home is translating debconf templates into Galician. I’m not doing bad at all. Although it’s just me working alone (by choice), mine is the 10th most complete translation, and climbing :) I’m going to document my workflow, for two reasons: one is that others may find it useful or may suggest improvements. The other is that, in this way, I won’t depend on my bash history not being erased :) I have a compendium with almost all strings in all debconf templates I translated, and several scripts I use to maintain that compendium, apply it to translations, etc. It is available at http://darcs.tarrio.org/gl-templates/, and can be downloaded with Darcs using this command:

$ darcs get http://darcs.tarrio.org/gl-templates/

In my local machine I keep it in $HOME/pos. (You may use this compendium and scripts freely, if you want, but be sure to change all appearances of “gl.po” into the appropriate file name for your language.) Nowadays I’m mostly driven by bubulle’s NMU announcements. Whenever he announces a NMU for a package I haven’t translated or whose Galician translation is outdated, I work on it. This is what I do to start a new translation for a package called “example":

1. Extract the attached templates.pot file into $HOME/pos/example_templates.pot.
2. Bootstrap the translation with my compendium, using this command:

   $ ./bootstrap-po example_templates.pot > gl.po

   The script is smart enough to know that a file called example_templates.pot is a templates file for a package called example. However, I can specify a different package name as its second argument:

   $ ./bootstrap-po nonsensicalfilename.pot example > gl.po

3. The generated gl.po file is UTF-8 encoded. However, I use xemacs to translate, which cannot read that, so I encode it into iso-8859-1:

   $ msgconv -t iso-8859-1 gl.po -o gll.po

4. Edit the file.
5. Encode the edited file into utf-8:

   $ msgconv -t utf-8 gll.po -o gl.po

6. Add the translations to my compendium:

   $ ./add-total-po

   (My compendium is the total.po.txt file, hence the name.)
7. Optional: edit the compendium to remove obsolete strings or strings which may be ambiguous so I don’t want them in the compendium.
8. Send the translated file to the package maintainer using the BTS:

   $ ./send-po example

9. Record and push the changes to the compendium:

   $ darcs record -a
   $ darcs push -a

This is how I update an existing, outdated translation:

1. Extract the file into $HOME/pos/gl.po.
2. Most likely, it’s UTF-8 encoded. Encode it into iso-8859-1:

   $ msgconv -t iso-8859-1 gl.po -o gll.po

3. Edit the file, fix only the fuzzy translations.
4. Encode it back into UTF-8:

   $ msgconv -t utf-8 gll.po -o gl.po

5. Update the translation using the existing translated strings and the compendium:

   $ ./refresh-po gl.po   msgconv -t iso-8859-1 -o gll.po

6. Open the file. Hopefully you’ll have new fuzzy translations. Edit the translation and complete it, or do some strings then refresh again and edit and refresh again until the file is completed. If the file contains lots of similar sentences you can just translate a couple of them, refresh, and the other ones will then be all fuzzy. Magic!
7. When I’m done, encode into UTF-8, update the compendium, send to maintainer, submit changes to compendium.

If you're a French person and were guessing what I have to do with the most famous French union, you've just been tricked. CGT stands for California Guitar Trio, a band of three virtuose guitar players. We attended their French gig yesterday night at 'Le New Morning', a famous jazz club in Paris. They came and played with Tony Levin who is, for the skilled music amateurs, one of the top world's bass player (*the* top for me). That was an amazing concert and great experience. Just try to imagine Pink Floyd's 'Echoes' played by 3 (electrified) acoustic guitar players, just after a 'Prelude and Fugue' by J.S. Bach, played cyclically (one note per player in round-robin order....a playing method invented by Robert Fripp and the League of Crafty Guitarists). That plus Bohemian Rhapsody, several pieces of their own and a stunning composition by Tony Levin where he proved that he *is* really a virtuose bass player..:-) I was just stunned, indeed....and, of course, also enjoyed a lot the presence in a very small club of Uncle Tony who I saw many times until now, but only in big arenas in all Peter Gabriel concerts I've attended since about 1980. CGT will play tonight (15th) on Milan, tomorrow in Roma, on 27th in Toronto, 31th in Rouyn-Noranda (Quebec) and a few times with friends during the Quebec Summer Festival. Then, they'll play in many places in California and all over USA. If you love eclectic music, just go listening to them. So, yes, CGT rocks!

27km...that's what Google Maps says, while my podometer says 28km (2/3 of a marathon). Anyway, I'm happy again as I pushed my limits slightly and ran that distance on Saturday. The former record was 24km. More happily, I could run another 20km today, 2 days later, and finished it very easily (the 27km run was much harder at the end but also had much steeper slopes). Long week-ends are good, indeed, to improve my running skills. For Dirk: the pace is currently 9:36 after converting things to your odd units...:)...so, I'm still far far away from your performances..:)

If you look at the po-debconf stats pages, you'll see that, as of today, the French translators translated 9693 strings out of 9688...:-) We are so clever! This is actually an annoyance on the scripts that generate these pages. Sometimes, in packages, the debian/po/*.po files contain *more* strings than the debian/po/templates.pot files. This happens when maintainers do not run "debconf-updatepo" in the clean target of their debian/rules files. People, there is a lintian warning for this...:) We will try to correct this (actually, more Nicolas Fran ois than me, indeed) but that will make my stats less easy to do... So, "sadly", we are indeed only about 99%... That said, the NMU campaign is still going on. Portuguese is now ahead of Czech (ranked 3rd). Russian passed Swedish and is very close to Spanish for rank 7. And, among the most actives, Finnish and Basque are still climbing up quickly.

Roberto Lumbreras reported bug #480000 on Wednesday May 7th. As bug #470000 was reported as of March 8th 2008, we're still keeping nearly exactly the pace of 2 months for 10,000 bugs, so 60,000bugs a year. Bug #500000 should then be reported around Sept 7th 2008. As a consequence, the candidates for winning the 500000th bug contest are still Miguel Gea or Kartik Mistry. Jacobo Tarrio has his chances, though. I haven't done any MBF to keep chances of winning the contest myself. See you around July 7th for bug #490000!

It might happen that you (I mean you, gentle reader of this blog) being a Debian developer or maintainer, just receive one of my numerous "Intent to NMU to fix pending l10n bugs" or "Announce of the upcoming NMU of ". The first gives you 10 days to react....then I send the latter which opens an 8 days translation update round. In such case, as the carefully written template for these (pretty long) mails says, please DON'T immediately upload. Speak to /me instead so that we coordinate the translation update round. That will also allow me to send you possible QA enhancements I found in your packaging while trying to build your package *and* translators to send more material. Some (rare) maintainers just uploaded immediately after getting these mails and they're now getting what could be expected: they're receiving new translations...and are then likely to be catched pretty soon by the patrol, again..:-). The "intent to NMU" mails are sent to the i18n mailing list and some translators find this to be a good opportunity to send translations in their language (that's why Finnish, Galician are for instance climbing up pretty quickly in statistics). The main intent of these NMUs is to get updates done, not doing NMUs so, of course, I will welcome uploads...it is however more efficient to coordinate them.

Last week-end, I sent several calls for translation updates for some key packages which I maintain localization for: apt, menu, noticeably. Also sent a reminder for Debian Installer beta2 even if Otavio already did so. That will continue over next weeks, of course along with the running l10n NMUs for which the pace is now about 2 packages per day. It's also good to see many package maintainers send call for l10n updates now in debian-i18n. One enhancement for the next release cycle will be avoiding to concentrate these during the last weeks/months of the release. Translators are not super(wo)men...:)

Karolin Seeger announced the release of the third pre-release of Samba 3.2.0 yesterday. That make a very short window for me to upload it to experimental (yes, experiental, this time) as I was planning to do. Still, reworking the Debian patches was easy so I hope I'll make it before leaving for a week on dialup. If that happens, please test as widely as possible. I think that it is safe to use Samba 3.2.0 for everybody running a machine with unstable (such as your own laptop or a non critical file and print server). Testing it will help upstream a lot.

A few news from the campaign of NMU's targeted at bringing po-debconf translation ratio as high as possible for as many languages as possible (and also improve packaging quality quite significantly on many loosely maintained packages).

• Galician translation are going up quickly. Each NMU features updates by Jacobo Tarrio and Galician is now ranked 10th in the translation ratio list. They could even hit the Spanish (Castillian) translation ratio and become the most translated language from Spain..:-)
• Finnish follows the same way. Starting from quite low, Esko Araj rvi bringed it to 13th rank and now above 50%
• Russian is also going up quite fast and could at least reach 70 or even 80%
• Portuguese (non Brazilian) hit the 90% mark and heads up to 100%
• German is very likely to be bringed to 100%. Crossed 97% today.
• French will be bringed to 100%, as it was for Etch. Close to 99% now.

Maintainers are incredibly cooperative (at least those who are active..:-)). l10n NMUs are always welcomed. Some prefer uploading themselves which is of course perfectly OK. I will now also focus my efforts to minimize uncoordinated changes to debconf templates. Those are generally motivated by the will to follow writing style guidelines, which is perfectly OK. I however still suggest maintainers to send call for translation updates *before* uploading packages with modified templates. Will probably soon post that suggestion to -release.

I spent the entire week at the SambaXP conference in G ttingen, Germany. The conference is the annual conference of the community of developers, contributors and users of Samba. I general call it "my annual german pilgrimage" as I attended all seven editions of the conference, the only FLOSS conference I attend on my work time, as my daily work involves quite a lot of uses of samba. This year featured a workshop or training session held by John Terpstra, longstanding FLOSS evangelist and member of the Samba Team for over 10 years, and Karolin Seeger, the brand new release manager of Samba. This has indeed been an incredible opportunity to have discussions with them about the packaging work for Samba. Actually, the work we did in the last 3 years to bring Samba Debian packages in a very good shape (when I don't screw up) is much appreciated. I think these days have been another opportunity to keep that link very closed. I have been impressed by the promising work of Karolin with respect to the preparation of the release and the very serious way she has to do that work....as well as her very friendly, while still discreet approach to technical discussion. Some work was done on Samba package bugs, though less than usually (the remaining ones are harder and harder to tackle!). I mostly work on Debian packages as well as .deb packages prepared by Sernet (the services company that organises the conference and employer of some Samba Team members), as these could some day become the packages provided by the Samba Team. Talks at the conference were pretty interesting, to keep connected with Samba 3.2 and 4.0 development. News from Andrew Tridgell about exchanges with Microsoft (yes, The Evil) and access to MS documentation, are very promising. The collaboration between Samba developers and Microsoft enginneers is now working well again, at the engineers level (as Tridge says: "lawyers are away, now, we can talk at the engineers level and restore the link that existed in the early 90's"). I also could measure the progress of the Openchange project whose ultimate aim is to provide a complete Microsoft Exchange replacement solution. They currently have working MAPI libraries and an Evolution plugin for Exchange is under development, while the bricks to build a server are patiently being put together. Good discussions with Julien Kerihuel, the lead (and French) developer and manager of the project. Jelmer Vernooij and I also settled the final plans to get Samba4 packages in Debian. I proposed to make a first upload to experimental, but quite soon to upload to unstable, the point being a largest as possible exposure of that code, so that upstream developers (Jelmer himself, Andrew Bartlett and a few others...) get as much feedback as possible. That upload will not be targeted for lenny (we'll block it from entering testing). For that reason and also because Samba 3 and 4 will certainly coexist for years, the source package will be named "samba4". Expect more news quite soon. And final conclusion of that week: I've also been delighted to be able to visit Alex and Meike in Hildesheim, as well as Andreas in Wernigerode. It's always a great pleasure to see longstanding Debian friends. At the beginning of the week, I planned to see *two DD and finally, at the end of the week, I can tell I've seen *three* Debian developers, finally!

Once again, I'm sitting in an airport (CDG) and enjoying the benefits of IP over DNS to bypass the outraegous rates for wireless access (about 10 euros for 1h if that didn't change since last year). Actually, having IP over DNS work is pretty simple. What you need is:

• A domain that you control (ie you can create en entry in it)
• A server running DNS services on the Internet, preferrably with a static IP address, preferrably with no need of external DNS services
• Appropriate software

"Appropriate software" here is, IMHO, the iodine package in Debian. There's nothing as easy to setup as it is. There is no package for etch, AFAIK, but backporting it is straightforward. You then need to create an entry in your foo.bar domain, such as "tunnel.foo.bar" and have it point to your static address. Then you run iodine on that server, listening to port 53 (with a configured password). That creates a virtual interfaces with something like 10.0.0.1 as address. When you need IP over DNS access with your roaming laptop, all you need is to connect to the expensive network, then launch iodine and provide it two parameters: the "tunnel.foo.bar" address and a local address (for instance 10.0.0.2). If everything works well, you'll be prompted for the password and voil . You have a tunnel between your local 10.0.0.2 address and the remote 10.0.0.1. The rest is a matter of network routing and you can now access the entire Internet through your remote server with no charge. The entire traffic goes through DNS requests to the local DNS server. That can be fairly slow but all a geek needs is receiving/sending mail and remote SSH access, right? This doesn't work in every place, unfortunately. Many wireless providers block such traffic in some way. But, at least in A roports de Paris, I can promise that will work..:-)

In mid-February, I started the lenny l10n NMU campaign, to tackle down pending debconf l10n bug reports. This page give the list of targets. I'm currently down to "isdnutils" in that list and, this week-end, I launched several NMU processes (a full NMU process lasts for 27 days). I'll now do my best to sustain a rhythm of 2 packages a day. Of course, several "NMU intent" mails end up with an upload by the package's regular maintainer, which is the best that can happen. I also fix many little QA issues in such process, most of the time trivial errors reported by Lintian (which was a lot improved recently, particularly about doc-base sections, menu sections, empty maintainer scripts and a lot of such things. As of now, 45 packages have been done (either by me or their maintainer) and 38 are in the pipe. As a result, all languages statistics are improving a lot. Four languages are now over 90% and 12 are over 50%!

fr   : 9507/9720,  213 missing, 97.80%
de   : 9349/9720,  371 missing, 96.18%
cs   : 8971/9720,  749 missing, 92.29%
pt   : 8757/9720,  963 missing, 90.09%
vi   : 7731/9720,  1989 missing, 79.53%
nl   : 7394/9720,  2326 missing, 76.06%
es   : 6562/9720,  3158 missing, 67.51%
sv   : 6547/9720,  3173 missing, 67.35%
ru   : 6213/9720,  3507 missing, 63.91%
gl   : 5870/9720,  3850 missing, 60.39%
ja   : 5836/9720,  3884 missing, 60.04%
pt_BR   : 5095/9720,  4625 missing, 52.41%
it   : 4538/9720,  5182 missing, 46.68%
fi   : 4589/9720,  5131 missing, 47.21%
ca   : 3954/9720,  5766 missing, 40.67%
eu   : 3942/9720,  5778 missing, 40.55%
da   : 3435/9720,  6285 missing, 35.33%
nb   : 3129/9720,  6591 missing, 32.19%
ml   : 2876/9720,  6844 missing, 29.58%
bg   : 2808/9720,  6912 missing, 28.88%
hu   : 2684/9720,  7036 missing, 27.61%
ro   : 2636/9720,  7084 missing, 27.11%
ta   : 2612/9720,  7108 missing, 26.87%
lt   : 2603/9720,  7117 missing, 26.77%
sk   : 2576/9720,  7144 missing, 26.50%
tr   : 2527/9720,  7193 missing, 25.99%

The lenny l10n NMU campaign is now running for about two months now. The objectives of this campaign are tackling pending po-debconf l10n bugs. The process is the following:

• Each day, pick two new packages from the list, going down and ignoring packages not in unstable
• for each, look over its pending l10n bugs *and* other easily fixable issues (of course RC stuff as well). I fixed many LSB headers bugs this way, helping out to achieve that lenny release goal
• try to build it
• if it builds successfully, send an "Intent to NMU" mail to the package maintainer and debian-i18n
• wait for 15 days (that leaves time to maintainers to react, or to have a discussion about the way to deal with such old bugs, etc)
• without news or in case the maintainer agrees before the deadline, send an announcement of upcoming NMU, again to the package maintainer and debian-i18n. Also notify translators with incomplete translations
• wait for 10 days
• build the package, test it and upload it to DELAYED/2-day
• send the full patch to the maintainer

Of course, when the maintainer wants to upload him herself, I immediately stop any action....but I still watch what happens and I nag her him in case nothing happens. In short, you can escape from my trap once I got you! Status as of now:

• 39 packages NMU'ed or updated by their maintainer
• 31 pakcages under work

A few utterly outdated and obsolete packages have also been removed fro mthe archive. All this helped the statistics for many languages to raise up. However, thanks to the effort of many maintainers who invest a lot of energy in emptying down the bottle we're filling up, the top languages statistics don't really move (new debconf templates, modified ones, etc.) With the lenny release becoming closer, I expect package maintainers to calm down now and leave us the opportunity to completee our l10n work without moving the target away when we reach it. The closer the release, the more aggressive l10n NMUs will be. Even for recent bug reports. Those translation issues are the easiest ones to fix, so please fix them. Please also note that I'll very happily sponsor any non DD upload which just fixes l10n issues.

A few days ago, I just screwed the samba packages maintenance. It's been quite some time since we were preparing Samba 3.2.0 packages. Upstream has not published a release version, only pre-releases. The release was planned for next week but will not happen as some last minutes issues have been found in pre-releases. Anyway, we had ready packages for 3.2.0 pre-release 2....which were planed to go to experimental. However, these packages were under preparation for quite a while, so when I finally agreed with Steve Langasek that it's OK to upload, I built and uploaded the packages. Then got a good laugh when discovering that I forgot "UNRELEASED" as target in debian.changelog. It was late, I then "s/UNRELEASED/unstable" as I often do in such case....then uploaded...then forgot about the issue and went to something else. The package went through NEW (it has a new binary package) and 1 or 2 days later quietly entered unstable. ...which I discovered only the next day, when it was way too late to remove from unstable something that was meant for experimental. All this explains why samba packages now use an epoch...:-) Things are back to normal now. Unstable has 1:3.0.28a-2 and experimental will hopefully have 3.2.0~pre2 soon. Next week I go to the Samba annual conference. I'll probably spend some time hunting down some bugs in the package and discuss with upstream about some of the issues we have (old bugs, FHS patch, upstream's .deb packages). And maybe also get better ideas about Samba 3.2.x schedule, to be able to decide with Steve and Noel which release is the target for lenny: 3.0.28* or 3.2.x.

Today, I left home at 08:05 to run in the local area. 1h40 later, I was back home after running about 17.5 (very muddy) kilometers. TV was on, Elizabeth was watching the Paris marathon. Runners had left at about 08:40 and were just reaching 20km. Depressing..:-)

There are things that are unique in FLOSS development:

Author: nekral-guest
Date: 2008-04-04 19:43:33 +0000 (Fri, 04 Apr 2008)
New Revision: 1966
Modified:
   debian/trunk/Makefile
Log:
Add a cheesy check.
Modified: debian/trunk/Makefile
===================================================================
--- debian/trunk/Makefile       2008-04-04 18:50:22 UTC (rev 1965)
+++ debian/trunk/Makefile       2008-04-04 19:43:33 UTC (rev 1966)
@@ -1,4 +1,16 @@
 PKG=shadow
 SITE=ftp://pkg-shadow.alioth.debian.org/pub/pkg-shadow/
+deb:: check_cheese
+
 include /usr/share/quilt/quilt.debbuild.mk
+
+check_cheese:
+       @dpkg-parsechangelog   grep -q "\* The \".*\" release\."     \
+               echo ""; \
+               echo " **                                  **"; \
+               echo " **  Warning: not a cheesy release!  **"; \
+               echo " **                                  **"; \
+               echo ""; \
+               exit 1; \
+        

Summary: releases of the Debian shadow package all have the name of a French AOC cheese since July 2005. Nicolas, who was focused on a new upstrram release of shadow inadvertently forgot to add such name for the 4.1.1-1 release and thus added a failure of the package build process in case on forgets the cheese name. We should however give NMUers instructions about the right way to NMU the package, such as giving the name of a local cheese.

That one could make the day of all maintainers who got REJECTED packages because of loose copyright information:

W: dak: copyright-without-copyright-notice
Maintainer: Joerg Jaspert 

No offense intended to Joerg, of course, who does tremendous work in the project...just some small pun..:-) Indeed, we have hundreds of packages in the archive which are missing that information, nearly always mixing *copyright* information with *licence* information. And the lintian warning is fairly recent, which explains that dak (who needs love.....the package in the archive doesn't contain the *real* dak scripts that are used on Debian machines) is reported with that warning.