Older blog entries for kov (starting at number 47)

Almost time to leave for debconf5... I hope everything works alright and that I'll be talking to fellow debianers by tomorrow 22:30, Finish time.

I plan to talk to lots of people, have lots of fun and have gksu in a better shape in unstable by the end of debconf. It'll be good to meet the friends I got to meet @debconf4 and to meet new ones, especially our DPL, who I longed to meet at FISL and unfortunately could not come =(.

See you in HEL!

Thanks! After a very cool party which I'm sure made some people angry because of the noise at FISL6.0, sarge was actually released and no one will make fun of us =D. lol

I know people are requesting at least a week to rest from all this and that some other people are already discussing wishlists.

It looks like another discussion has already begun at debian-devel about the Vancouver proposal, too, and I've seen lots of positive comments and attitudes from some people (porters, even) saying that they're interested in discussing the proposal and fix its bugs instead of simply throwing flames. It looks like many people understands the proposal as a 'we only want to support arches a, b and c'.

I'd really like to see people cooperating on fixing eventual problems on the Vancouver proposal. It seems clear to me that we do have problems with our porting infra-structure and that some stuff are causing problems and delays. Making sure our day-to-day work has got quality is perhaps more important than guaranteeing the quality of the results alone, and that's the main goal I see on the proposal.

Another important thing I think we should be considering seriously is on setting a release date. A release date helps in a number of ways. You can plan what are the main transitions you want, what main version upgrades you want, etc. You also replace the feeling 'we won't be releasing anytime soon, I can delay this a bit' for 'we will be releasing in 9 months, I better get what is important for me ready or at least started'. You also end the 'maybe we have some 2 more months, let's try to push a new version of $stuff in?' dilema.

Just as an example, if we knew we would be releasing on June 20005 the GNOME Team could probably have targeted GNOME2.10 as the target GNOME version for sarge. The team would not have done the enourmous work it did on GNOME2.8 while 2.10 was in preparation; focusing on GNOME2.10 even before it was released we would have it ready on time.

My own pet project, gksu, would have had translation updates sometime ago if I knew when we would be releasing. The 'we won't be releasing anytime soon' feeling made me delay this 'cause I had more interesting stuff to do - until the release target of may 31th was announced. Then I tried to do all translations updates and some simply didn't have time to come in.

Now, if we know we're going to be releasing on June 2006 we know we can target GNOME 2.14, which should be released around march that year. We can start packaging GNOME 2.13 on experimental as soon as it comes into existance and move it to unstable and finish its debuging process sanely.

I'm sure we can come up with other good advantages this would have. ABI transition planning, not increasing RC bugs by not adding new upstream versions close to the end of the release cycle, not making our future stable too old for not blocking development because 'we can freeze anytime'.

I'd really like to see this being considered seriously and discussed with sane arguments instead of flames =).

So we've watched these last days the announcement of the proposal of the creation of a new organization in Brasil. It's name is 'OMSL' - Organização Mundial do Software Livre (or World Free Software Organization), a quite pretensious name. The goal they state publicly is to:

offer a concrete response in orchestration among countries, public and private companies, universities and research institutes with the goal of organizing the participation of every potential party in the new software market (my own translation of text found at http://www.omslnet.org/html/o_que_e_omsl.html -- yes, the World Organization does not have an internationalized page yet).

Their executive summary (portuguese only) states lots of what they want to do and this includes certification programs, inclusion of traditional methodologies in the FS development process and some ways of choosing and standardizing the 'best of breed' solutions in the FS world.

I belive that, as a world organization-wannabe, OMSL people will want to receive input from people from all the world on how they're starting, so maybe you'll want to send your comments to osml@omslnet.org.

On other news, yesterday I uploaded the version I consider to be final of all the gksu-related packages. The upload intent was mainly to update translations, and I indeed ended up uploading three new translations and some translation updates for all of the packages.

Now I'll think about doing the same for APT-HOWTO.

So I had a very good time at my hometown this weekend. But before talking about this let me acknowledge two things: so Branden was the winner and I'm very happy about this. Having voted him as first choice for the two past years and as second choice this year I'm very confident he'll be able to help Debian get back to being a Project which we all love and have fun working on. I am very happy with the news that came from debconf5 people that we got loads of sponsorship and will thus be able to fund everyone who requested funding. More than anything, though, I would like to emphasize something Andreas said on his message:

"I think we need to re-calibrate how we think of Debian ourself and remind us that we managed to create an enormously valuable system. Now lets get our act together and lets become the kick-ass Über-distro that rules the world! Yay!" (Andreas Schuldei on debian-devel-announce)

About the weekend, we had the third Linuxchix Brasil meeting this year and it was held on my home town, organized mainly by our friends Priscilla and Caroll with the help from Sulamita and Lullys from São Paulo. Thanks, girls! It was definetely amazing! It was a very nice time meeting people, giving a course on Subversion with not a lots of materials but with some nice 'walk-through' with people doing real editing stuff and a talk on GTK+ followed by one on QT by Helio Chissini, who has some nice stuff on his blog. Btw, I love pyslide.

On the hacking front, I've been messing around with reimplementing a 'Debian Brasil Counter'. We used to have one I coded in a very bad-written PHP some years ago and I've decided to rewrite it using python (cherrypy+cheetah). It's going cool. Check out its svn repository.

29 Mar 2005 (updated 29 Mar 2005 at 00:33 UTC) »

I noticed I forgot to mention Matthew Garrett on my last post. Well, I like him and think he would be a good DPL. His proposals are good but I think he is wrong in the way to fix communication and to achieve consensus. I think we need stronger solutions for the problem.

As andrelop mentioned decko is working on packaging update-{notifier,manager}. We had a problem, though: update-manager only works with python2.4. Discussing this in #gnome-debian we came to the conclusion the best thing to do would be porting it so it would use python2.3, which is what pygtk is ready to support on Debian for now. I then patched it and it works.

Now the problem is Debian's python-apt lacks wrappers for DepCache stuff on libapt-pkg, which Michael Vogt developed for ubuntu and used on update-notifier. I talked to Matt Zimmerman and he said he'd upload a new python-apt version including this to experimental soon for us to use it, though. Thanks guys!

One thing I was thinking today was: what happened to all Progeny cool ideas? I watched their presentations on picax and componentized linux but never really heard of them again. Ubuntu has been able to echo much more 'till now, but I would really love to see the ideas guys from Progeny brought to Debconf4 being tried out. Especially, I'm sad we still could not integrate the python configlets a bit more on Debian. Now that we're working on bringin some ubuntu tools and even yast for Debian, maybe the Debian Desktop project should do some effort on bringing the configlets together for etch? =)

I definetely agree with Andrew Pollock's post: screen rocks my world!

I and my coworker coredump always have screen sessions running on the servers we administer together, and are always looking at logs and fixing problems together. This was also how we hacked the scripts that we used to convert all our users' mbox-based mailboxes to maildir some months ago when we fixed the mail server of the Ministry

I'm simply unable to imagine myself living in a world without screen.

About the vote... my ballot:

[ 6 ] Choice 1: Jonathan Walther 
[ 3 ] Choice 2: Matthew Garrett 
[ 2 ] Choice 3: Branden Robinson 
[ 1 ] Choice 4: Anthony Towns 
[ 4 ] Choice 5: Angus Lees 
[ 2 ] Choice 6: Andreas Schuldei
[ 5 ] Choice 7: None Of The Above

I was quite surprised by Anthony Towns running for the job this year. Although I voted for Branden twice now I decided Anthony Towns seems to be the one with the most concrete proposals on how to get Debian back to being a sane environment which is fun to work on. I still trust Branden would do lots of good on this matter, too, if elected. I decided to give the same trust to Andreas, too... while I think he'd not be as tough in some respects as Branden would I think he'd get results by using other strategies.

I don't know Angus a lot but he seemed to be sane enough to be ranked 4. Jonathan on the other hand has demonstrated that he does not deserve to be trusted and that he is able to completely change his "image" and opinions depending on convenience. I don't want someone like him as a DPL, ever.

Like Robert and Jose I had problems with the fontconfig upgrade, although I don't usually have them. I felt like living one year ago or so again when I saw this. Then I went back to fontconfig 2.2.3 to get this back. Someone said something like "why did it disable the autohinter?" on #gnome-debian when I showed the screenshots. My Bitstream Vera fonts are being rendered like crap, it seems. I know little about all this but will try to dig a bit into Robert's observations tomorrow morning.

16 Mar 2005 (updated 16 Mar 2005 at 23:25 UTC) »

In my last post I talked a bit about how I was enjoying to feel that Debian was understanding that "when we have problems we fix them with Really Good Work, not with long flamewars which end up seeming to just want to protect some kind of holy status quo".

Well, I think Anthony Towns expresses this idea in a practical and very easy to understand way.

Just finished reading the notes from the release team meeting a while ago. Controversial, for sure.

What I think about this is this is a great step on making Debian live again. I've been a developer since 2001 and I felt the same as Adrian Bunk many times... we like to see our work being rewarded, and one of the ways to feel rewarded is seeing your work being used on the Real World. This bad feeling surely made my motivation go down a lot.

OK, so we're droping support for some arches... in this regard what I think is we have very high expectations on the results but really lack comparably high expectations on the day-to-day work. Some time ago the GNOME packages were really badly shaped. They were, IMO because of exactly this reason, only marginally integrated into the system as a whole.

Then came the GNOME Team and proved that GNOME could become a very well packaged and maintained desktop inside Debian. The GNOME Team was good enough to raise Release Team's trustness in that it would respond fast and well enough to breakage and transitions, so GNOME 2.6 was allowed into testing, then 2.8...

The same then happened for KDE and to the Release Team itself through the time. Those teams make information flow high, so motivation and rewards are high as well. What I see is Debian understanding that when we have problems we fix them with Really Good Work, not with long flamewars which end up seeming to just want to protect some kind of holy status quo.

If you want your arch/package/anything considered for release, then get it ready and high-quality enough that it will be accepted in time. Show us the code. Great work! I feel my motivation regaining strength! Go, Debian, go!

38 older entries...

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!