Older blog entries for DV (starting at number 189)

Free Software and diversity

There have been lot of fuss recently around the GNOME project concerning how such a project should be run, who participates, etc. Of course there have been the mess about Eugenia article, i.e. how the project should get input and make technical decisions (trying to avoid the most inflamatory part and focuse on the interesting aspects), but to me what is really important is who is making the project. To me open-source is a process, but an open-source (or free software it's equivalent there) project is a set of people, and the project is the sum of the people activities. That's why I found Anne Ƙstergaard post on the foundation list more interesting, that's also why I'm really concerned that on the GNOME worldwide map there is only:

  • one person for the whole Africa and Mddle-East
  • two persons for China-Japan-Korea

We need to understand why we are not an attractive project for those part of the population. For women I can't evaluate why it is so, except I followed the "classical" Math/Physic and later Computer Science studies track here and as it was getting closer to core CS stuff the ratio of women student decreased from 50% general studies, 25% in Maths/Physics, 10% in CS engineering, 5% for CS doctorate, and a couple per cent apparently in OSS projects. Seems women have a hard time taking this geek stuff seriously, but when they do, well they do it very well. Maybe the key to get women involved is to prove it does have an effect and reduce the visible part of flamewar and pissing contest around (i.e. grow up !).

For Africa/Middle-East/Asia involvement, I don't understand well either what is missing, first it's so diverse that it certainly can't be reduced to a single cause. I'm really worried that we don't have one korean person showing on that map, this is one of the country where the Internet is the most integrated in the society, Linux has a serious deployment there, why is there no GNOMEs ? My main fear would be that we are showing a cultural bias upfront, and what can we do to solve this ? Should www.gnome.org be translated in a variety of languages with localized subsite ? According to the translation statistics, GNOME is 100% transalted in korean, so obviously there are people behind that, but that community does not show up. Why ?

I'm afraid I have only question but no real answers, honnestly I would prefer we spent energy trying to understand why we don't grow the community, rather than growing the difference in opinions within the existing community out of proportion (which is where I think Eugenia's attitude is really really wrong).

Sometime users are obnoxious too, I got a fantastic example on bugzilla this week (no link, no :-) , usually it is still worth trying to discuss even if it seems to go nowhere, I think that's how one can grow long term respect.

The 2.10 release

Was clearly a success, there have been some hiccups, but we are improving over time, this is very comforting and congratulation to everybody involved !

The ISO CDs were as far as I can tell based on my server acccesses a really really good idea, this need to be integrated as part of the release process if possible so that luis can sleep a bit the weeks during releases :-) !

Furious w.r.t. the Software Patent handling of the EC

Clearly we are all getting endangered by this. It also show how little consideration we get from the European Commision. I wrote emails to EU elected representatives, people I voted for, never got an answeri even automated :-(. I used to be enthusiatic about the european construction, in practice I'm afraid it allows just politicians to dilute their responsabilities, and then they are in heaven to take advantage of their position without being at risk legally. I think I will end-up voting against the European constitution here, and really really I would not have expected this a few months ago.

In the list of industrial "partners" who can now forget getting direct help from me about libxml2 or other open source projects, there is a nice list at EICTA:

07/03/2005: EICTA welcomed today's decision by the EU Competitiveness Council to adopt a Common Position
on the proposed Directive on the patentability of computer-implemented inventions.

This list some of our "friends" like IBM, Sun Microsystems, Adobe, Alcatel, Bull, Apple, Cisco, Canon, Dell, HP, Fujitsu, Motorola, Nec, Phillips, SAP, Siemens, Thomson, Toshiba. I'm in a positively furious mood at the moment and looking for possible retaliation (yes I'm really unhappy). So what can I do to express my miscontentment, and tell those "allies" that I would like them to play by the rule. One of the thing I can do is send them a nice mail each time that anyone from those companies asks for help about some of my code, telling them that I can only afford to give help in the framework of a legally binding contract with my corporate employeer. Anyway I doubt an isolated action is likely to help anybody, only concerted wide scale and clearly directed action might help. What would be the effect if 80% of the open-source projects maintainers would just refuse to help those company outside of a paying contract.

Anyway I don't expect a large scale revolt agaisnt this though I would like to be surprised, I think I will just set-up a web page on veillard.com explaining why I won't give free support anymore to some companies, and just use that as a canonical answer to mail coming from them. This won't block people from asking from a personal email for example if they wish.

the foot trademark issue

As luis pointed I asked the board to get the right to print T-Shirt for computer events. I made the request in a sufficiently scoped fashion that I think the lawyers should not oppose this. Then if this go though I want to expose this wording so others can get the same kind of approval for local geek show. But I don't think I flamed the board (I would have to flame myself, ouch !), I'm just trying to get a process which others can follow to achive GNOME marketing without making risk on the trademark.


FOSDEM was huge, as expected, a lot of things going on in parallel, I could not talk to all the people I intended to chat with :-\, still managed to listen to some nice talks, make a talk on libxml2/libxslt and a speed talk on gamin. I need to put them on-line and update veillard.com Talks section.

23 Feb 2005 (updated 23 Feb 2005 at 23:45 UTC) »


Going there with Dodji this week-end, I will do a talk on libxml2/libxslt in the Gnome room, and a speed talk on Gamin. The conference was already large last year but I'm hearing lot of people saying they are coming, I wonder how huge it is gonna be, I feel something absolutely crazy coming...

World Domination ?

In correlation with the growing size of FOSDEM I find interesting to see that the Gnome World map seems to balance heavilly toward Europe, while people used to oppose KDE/GNOME as being Europe/USA. But I must admit I feel frustrated seeing noone from the whole African continent, nor anybody from the China area (except William "bill" Brack which registered with his Hong-Kong coordinates.)

Is the European GNOME growth a side effect of GUADEC the main GNOME conference being done in Europe years after years ? What would it take to make conferences in China (volunteers, time and money :-\). How much companies beneficiting from Open Source are ready to spend to raise awareness of our projects in Asia and Africa. I wonder what would be the result of a few dedicated people doing in Beijing what the FOSDEM team did in Belgium for a few years. Probably something massive and an incredible return on a relatively low investment, who has the time, vision and money for this ?

Never underestimate the amount of work needed to build one of those conferences too, congratulations to all involved worldwide !


After annoying GNOME sysdamins and stealing the html and css code from the Evince project, we now have an initial Sabayon web site, with mailing list, bugzilla and basic informations to get people started if they want to play or help with the project. There is quite a lot to do, and testing is not trivial since we depend on very recent (nearly CVS head) versions of pyGnome, gamin-python, etc, check the README file ! Simplest is to use the RPMs.

One of the main thing we need from our target audience i.e. people doing sysadmin setups/maintainance is feedback on some of the design Issues.

Software Patents

Anybody able to read french should read the on-line article from Le Monde where Michel Rocard (ex french Prime Minister, socialist) explains how much battle and pressure occured under the hood at the the European commision on software patents. To borrow a short translation found from a commentator on /.:

We never could have talked a common language with the companies representatives we met - in particular those from Microsoft. Speaking about free ideas circulation, free access to knowledge, was like speaking chinese to them. In their way of thinking, everything that is not usable for immediate profit cease to be a growth vector. They don't seem to be able to understand that an invention which is a pure spirit creation can't be pattented. It's simply terrifying. Many of us, at the Parliament, agree to say that they never have know such a pressure and such a verbal violence during their parliamentary work. It is a huge case.

Other translations from me that time:

The Parliament vote, the 24th of September 2003, adopted by 361 vote for, 157 against and 28 abstain of amendments [restricting the field of software patentability]. This caused the furor of the commission and led to a storm of mails coming from 15 large groups led by Microsoft.

We learned that to draft the first project, the Commission used the work of external experts because this was a new topic to them. But many of those external experts were coming from Microssoft and other software companies.

What IMHO make this big, is that it's such a failure and there is so much griefs that someone with a very high position like Rocard ends up exposing the case himself, publicly and to the main newspaper in France !


I didn't blogged for 2 months, but I did some code anyway, I just released version 0.0.24 of gamin which should fix the damn Desktop update problem people on Fedora Core 3 have been complaining about. I also worked on libxml2 for example working on XPath streaming subset but this is not finished, it's in part used by Kasiemir to finish the implementation of XML Schemas constraints.


Okay I also worked on sabayon, mostly by giving its name in a 2mn IRC challenge with Mark, implementing python bindings for gamin, and working on the RPM packages. You can visit us on #sabayon on gnome IRC if needed, but best is to use the RPM packages, it requires very recent versions of various packages. We also lack a mailing-list (asked) and a web site :-).

Internet Radio

I have been pointed to Radio Paradise Internet radio, and have been listening to it from xmms the whole day, truely excellent, I'm tempted to make a donation since they re listener supported (no ads !).


Seems I will be back on the board again next year. A bit disapointed by the low number of overall casted votes.

Mapping file in partitions

Old-timers like me probably remember the PCTools defragger interface where you could see the map of the blocks moving while data were compacted. I had the need to at least get an equivalent static picture for a set of files in Linux partitions. Shortest way was
1/ a small program extracting filesystem and data index informations and generating XML data, this need to run as root. It takes the list of files either on the CLI or using a --list filelist option
2/ a small XSLT stylesheet generating SVGs out of it (there is a line wrapping bug I know, not implemented yet).

First example was to map the set of file from /etc/readahead.files, here is the first map. Then by doing:

mkdir /temp
for f in `cat /etc/readahead.files` ; do \cp -f $f . ; done
rm -f temp/*
for f in `cat /etc/readahead.files` ; do \cp -f $f . ; done

The new set of files in /temp looks easier to fetch from disk. So at least from a basic experiment it seems one could do some interesting trick to speed up loading a bunch of files even without implementing hard kernel APIs. This still need a lot of work to really get an estimate of how best to do this and how efficient it would really be (plus the map does not include the directories and fs metadata)

BTW is there an SVG viewer with <title> support, i.e. showing a tooltip with the string associated with the element, that would allow to just point at the block and see what file it maps to, that would be extremely convenient and possibly fun too. Speaking of fun I ran it for the list generated by rpm -ql openoffice.org ... try it, it's scary ! The fact that I ran multiple upgrades of my system instead of fresh reinstalls is likely to have made thing worse.


Not much to say except :

paphio:~/XML -> ./testapi -q
xmlRegisterCharEncodingHandler: NULL handler !
Total: 1092 functions, 267029 tests, 0 errors
paphio:~/XML ->

I strongly suggest people upgrade if possible .

Gnome Board

So I'm running again after a one year break. Seems to me the next step is to grow the GNOME project awareness, build local representations in countries outside the US, try to federate them in a loose way, and ramp up the propaganda machine. I'm not sure I'm really the best person for this but apparently most candidates share the same goal, and I can still take minutes, and work on the web site.

Hardening libxml2

I have spent last week focusing on a single task: hardening libxml2 code by the mean of autogenerated regression tests. Though the principle is extremely simple: for every type in the public API then build generators and destructors for those types, then for every entry in the public API generate a set of calls for all the autogenerated values, then run the resulting program tracking illegal memory accesses and leaks. It has proved extremely efficient raising tons of issues in not frequently used libxml2 API. This kept me really busy for the last 10 days, extending the set of API and types covered, running gdb, valgrind, and testing on 64bits. I don't know if I should be happy by nearly finishing this very boring but needed task or ashamed by the number of issues found.

The current version of the tests covers approximately 1100 libxml2 entry points and generates more than 250,000 test calls to the API. Once I think I'm done with it I will release a 2.6.16 version.

Land of confusion

I positively dislike when people, by lack of grasping a concept just fall back to using the words without trying to understand them. When this is used to make propaganda then I get really annoyed. When this is used to attack something I believe in using the disparity between the word and its meaning then I really want to bite. Sun's CEO has been telling "Solaris is our implementation of Linux" for years, not just recently. It always made me angry. To me, Linux is a *process*, successful Open Source is a process. An Open Source Licence is a necessary condition to develop an Open Source process, but this is *not* what makes a software have the properties that are usually praised in Open Source development. The history is now full of example of code which got switched to an Open Source licence, and failed as an Open Source project. Adopting an Open Source Licence is a necessary first step, but not sufficient. Growing a community, distributing the decision process among that community, making that process fully transparent are keys to understanding what the words really mean behind "Open Source Success". I also strongly suggest people read Linus write up on Linux kernel management style, yes I consider this a MUST READ, it applies globally to any well managed Open Source project, and it's even fun to read !


There are still some nasty bugs embedded in gamin. First make sure you use a recent version, 0.0.14 fixed a big problem. I got a good bug report where gamin would deadlock client and server on reconnect, nasty but at least it's now understood. Also check the debugging page. I'm looking forward discussing with rml In Boston this week-end. I'm seeing more and more reports from people using the inotify back-end, this is good.

Boston Botanical Garden

I lived in Boston/Cambridge for 18 months, and while I'm quite fond of botanic and especially orchids I do not remember seing any botanical garden. Apparently there is a new project to build a big Botanical Garden between center downtown and the seaPort. Looks interesting but not yet ready.

Sleep troubles

I have had serious sleep trouble for the last week or so, I just can't get asleeep before 2-4am, I hope I will be able to sleep in the plane and that the timezone change will allow me to get my sleep back in shape.

180 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!