Recently in Informatique Category

So the last afternoon @ fosdem 2008 I ended up listening to a very informative talk on autoconf, automake and libtools. The argumnt was that they were the tools of the trade - and that it wasn't so complex. The rest was more a walkthrough and at the end I was not reall convinced that those tools where not easy nor complex to use properly. The reason I ended up in this metting was that the presentation on making robots with 8 bits controllers and linux was so crowed that there was a waiting line in front of the room.
Then I went back to the mozilla room to hear the plans that mozilla had for bringing a browser on mobile devices. The talk draw lots of attention and the room got more people. Max jumped on his chair when the speaker evoked the idea or removing Mathml
from the mobile version of the browser.
Once agan fosdem was very nice and I'll probably attend next year.

Today I started uploading some pictures from saturday. Then I went to Janson to listen to Hudson by Kohsuke Kawaguchi.
Hudson is a build automator aimed principally at building java apps. It seems the main interest in that project is that you have a nice UI for you build tool and that it's easy to administer through the web GUI, other than that I did not really find anything very interesting on the product itself.
Then I went to meet my friend wolfgang in the OpenSuse room to follow a very interesting talk on the energystar label. I learned that to get the label you need to have the hardware and the software to be correctly setup. The levels of energy that devices can waste and still get the label is pretty high. The talk was mild-technical but complete and yeah interesting.
CMake presentation by one of it's author was nice - but missed one point in the very beginning : what is cmake is about - which I got just a few minutes later. The presentation looked more like a workshop because we where presented with examples and how they worked or were supposed to work. Nice things are ctest and cpack , to unit test and make code coverage and then package the product you are building.

Sitting in the OpenSuse room now listening to a talk on how to package for that version of Linux - the guy talking has a strong german accent, changes from the french one :-).
I've been sitting in the mozilla room for most of the day - the room was smaller than the previous years and that was a pitty because attendance was much higher. It's nice to see people that I have not seen since last year, it's like friend with a common passion - that is what empowers free software communities. The room was so hot that someone came up with a nice joke on the blackboard "sauna.fosdem.org" :-)
The talked started by a state of the mozilla community and software distribution, with nice numbers (53% of downloads are english ones), and information on how firefox beta 3 was used (a very nice graph) and what the plans for the near future might be.

So we had two very nice presentations on the calendar project behind hosted by mozilla, with many questions and usual answer : we need more developing man power. Gave a nice idea of the state of the products and the dev team.

Previous to that Dan Mills talked about weave a project hosted on labs.mozilla.org. Weave is about sharing profiles, personnal information and bookmarks. I like the idea - and the concept they are working on, it's a bit like having del.icio.us and other services mashed up together but with privacy options. At the end two technologies were pointed to Dan Mills, one is p3p and the other one is foaf.

The next session was on Thunderbird and what shold happen this year interms of releases and staffing for the newly announced mozilla-messaging corporation. Very nice and very interesting especially if we get Sundbird inetgration by the end of the year.

I'm attending FOSDEM again this year and already met a few of my fosdem friend, one of them just reminded me that this was the 5th edition of fosdem I was attending. I'm sitting in chavanne, the biggest room (around 1,5 k places) listening to a "How a large scale project works" by Robert Watson from the FreeBSD project. The talk is not really technical but gave a few interesting thing about the social aspect surounding the development of the operating system.
The previous presentation was on the use of linux by the film industry and all the studios use linux in render farms and also on the desktop, some of the visual effect from known films where shown - that part was very interesting, but the rest of the talk was actually quite boring. So seeing what was added and not noticing it was really nice. Also learned that each frame is rendered independantly from all other frames.

A bit more than a year ago I was explaining how I managed our test cases for joost. Things did not change much since then, but are on the verge of changing. After my trip to mozilla, I did spend some time investigating Litmus but found that it was really too tied to bugzilla and did not felt like hacking much on it. I did like the fact that Litmus is written in perl. Since then I we got experienced QA people that joined joost and the first thing they did was search for a tool to use for storing and running test plans. The chosen solution is named Aptest. I've had the pleasure of installing aptest, which is also written in perl. For now I can't really say that we are using it, because we need to spend some time figuring the application. I'll keep you posted on our usage and our tools.

Je viens de lire que l'organisateur de la session de signature de clef numérique était prêt. Cette année je n'y participerais pas - en effet j'y ai participé les années précédentes en me disant que le meilleur moyen de combattre les pourriels était de signer tout ceux que j'envoyais. Mais cela ne fonctionne pas - les autres utilisateurs ne comprennent pas ce que la signature représente. Je ne vais donc pas perdre trop de temps avec les signatures cette année.

Just found about http://upcoming.yahoo.com. I like the concept - looking for events, entering events so you can tell the world about thing you like attend. There seems to be export fucntions for ical, yahoo calendar and google calendar, which is probably the type of service I was looking for without searching specifically for it. The other neat feature I like about it is that I can tag my flickr pictures with a machine tag and then the picture becomes relevant to the event. As I'm attending fosdem this upcoming week-end, I registered the event (could not find anything by searching for Fosdem, and for searching events in Brussels). After searching by end (and not querying around for known names), I was able to find the registered fosdem events. I deleted the one I had created. I just hate services where search does not work.
I need to explore the service a bit more and see if it can be a bit more useful.

As I reported last year the 2007 edition of that camp was a lot of fun - at least the only day I attended. The 2008 edition as been announced, all the details are here. The program - which is not finished yet - is available here. I'm probably going to attend and take a bit more pictures than last year.

index.rdf et index.xml ne sont plus mis à jour depuis que je suis passé à Movabletype 4.x, merci de mettre vos lecteurs rss à jour en les pointants vers atom.xml.

I kind of like the idea of not connecting to ebay at all for a week. Relaying the message here as I will do that myself.

Yesterday I passed a simple command :

svn remove XXX
svn commit


The directory I was trying to remove was 3GB in size. It took svn 1/2 GB of real memory and 3GB of VM. The operation took 1,5 hour to complete. The good news is that I don't intend to pass such command every five minutes.

Et jusqu'à présent l'outil est conçu par l'homme pour répondre à des besoins et on adapte l'outil à l'environnement dans lequel on compte l'utiliser. C'est mon ami Max qui a lancé le débat la semaine dernière alors que nous accomplissions des formalités administratives, sur l'accentuation des majuscules. Et dans la foulé il remet une couche sur le O e dans l'o, un sujet que j'avais abordé. Je suis d'accord avec Max sur le fait qu'il faille agir maintenant, mais je ne penses pas qu'écrire à son député suffise ou au ministère de l'éducation nationale suffise, ce qu'il faudrait à la langue française c'est l'équivalent de taalunie pour le néerlandais, une commission composée de membres des différents pays utilisant la langue de Molière comme vecteur de communication afin de veiller à l'uniformité d'icelle.
Mais je digresse - je disais donc que l'informatique est un outil et l'outil ne devrais pas forcer un changement lors de son utilisation mais devrais être adapté à la tache pour lequel il est conçu, on ne change pas les pies d'une vache pour faciliter la tache des constructeurs de trayeuses. Or il se trouve que de plus en plus de documents sont écris à l'aide de logiciel dit de traitement de texte. Ceux-ci sont proposent souvent une fonction pour corriger l'orthographe et ces fonctions de signalent jamais coeur comme étant une erreur. Donc ce que je propose en plus des lettres aux députés, sénateurs et ministères c'est de faire pression sur les éditeurs de correcteurs orthographiques afin que ceux-ci considèrent les oublis de œ comme de fautes. La liste des gens à contacter est la suivante :

• Microsoft


• Apple


• Hunspell


• Ispell


• Corel

Et je serais très curieux de avoir comment se comportent des produits plus spécialisé comme "le correcteur 101" ou "Antidote" sur ces deux sujets.

Reste ensuite ce que j'appellerais une hygiène de vie : se forcer à écrire proprement. Ce qui se complique pour les expatriés que nous sommes perdu avec des claviers qwerty - il faut déjà se battre pour ajouter les accents, on ajoutera donc les accents sur les majuscules et les œ et æ et les techniques pour le faire sont décrites ici

J'ai jamais été un fan d'orthographe - j'ai vraiment le sentiment d'être un vieux con réactionnaire en écrivant ce billet. Mais j'ai vraiment le sentiment que ma langue est une partie intégrante de ma culture et donc de moi même. En forçant ce genre de petit détail je tente sans doute de faire mon Astérix.

Je viens d'ouvrir un radar chez Apple afin que le problème soit pris en compte - on dit que l'espoir fait vivre - ben je crois que sur ce coup là je vais vivre très longtemps.

Voila je viens de créer mon fichier foaf.rdf. Foaf veut dire friend of a friend - littéralement "ami d'un ami" . Je l'ai créé en utilisant l'excellent foaf a matic.
Pourquoi m'y suis-je mis ?
Ce fichier contient les liens sociaux que j'entretient sur le web ( toute personne ayant un compte sur livejournal ou bien advogato par exemple ont déja ce type de descriptif social). Et bien je passe du temps sur facebook (pour jouer), sur flickr, copain d'avant etc ... A chaque fois je dois rentrer les mêmes informations (qui je suis et qui je connais, quelle type de relation j'entretient et avec qui) et je n'en suis pas le propriétaire, je ne les contrôle pas. Avec le projet foaf je contrôle ces informations et j'ai bien espoir que de plus en plus de site web se mettent à lire ce type de fichier ). L'autre très bonne nouvelle est le fait que google indexe désormais les fichiers foaf.

The Library of congress uploaded around around 3k photos on flickr. The idea behind that is to help the library catalogue it's picture. The deal seems quite simple and not that time consuming , it's about adding tag to the pictures. I like the idea which is described in details here. So I've spend some times adding a few tags to the pictures present in the photostream.

I'll be attending fosdem this year. I'll try not to stay in the mozilla room for the complete week-end.

So I now own a new machine for a bit more than a week. I've installed leopard and decided to go with mac ports to give it a try instead of using fink.

I kind like mac ports and for the usage I have of such tools, using fink or mac ports doesn't really make a difference. Since I now have a decent machine I decided to have a look and try to participate to some nice open source project, as my new hobby is photography - I've figured I could maybe be useful somewhere in that area. So I started having a few peaks at libopenraw and exempi which are being developed by a friend of mine. The project aims to bring proper support for the raw file format used by camera makers to linux.

So after installing the dev tools and trying to configure and compile, I was able to achieve it for exempi after hub updated the autoconf/automake files and scripts. This helped configure not to fail so early for libopenraw. It now fails with :

checking for jpeglib.h... yes

./configure: line 20023: syntax error near unexpected token `EXEMPI,'

./configure: line 20023: `PKG_CHECK_MODULES(EXEMPI, exempi-2.0 >=

$EXEMPI_REQUIRED)'

Googling does not really help. What I did though was to use another machine which runs 10.4 instead of 10.5 and uses Fink to manage dependencies instead of Mac ports. On that machine I had no issues due to tooling - a few issues at compile time but they where solved. So I now need to figure out if my issues come from 10.5 or mac ports - good thing I get some vacation time next week.

Yesterday I realized I had fucked up my dns data , and wanted to ssh to my box which gave me :

percent_expand: NULL replacement

Which according to google is macports issue :-( I got on my box using my older laptop (the one running 10.3) and fixed the issue, but I'm being annoyed by those issues I'm encountering.

Last issue I had was getting pictures out of my Nokia 73. Launched the bluetooth browser and was unable to connect to my phone. Killed the app and tried to pair the phone and the machine which failed (console.app is empty ) so I might retry tonight after a reboot to see if the first kill didn't leave the bluetooth hardware in a unknown state that prevented the pairing afterwards.

Only thing I'm not complaining too much is civ IV (except that I don't see the point of having 3D graphics for that game, it does not bring much to it.)

Hier j'ai passé une bonne heure à mettre à jour le firmware de mon N73. J'avais déjà eut la velléité de le faire, lorsque j'étais entrée en possession du téléphone. Pour réaliser une mise à jour, il faut :

• Une carte mémoire


• Le cable de chargement


• Un Pc


• du temps

J'ai décidé de mettre le téléphone à jour suite à ce billet sur le groupe n73. Cette fois le téléphone était équipé d'une carte mémoire celle-ci est indispensable pour faire une sauvegarde.
Ensuite le logiciel de mise à jour proteste quand le téléphone n'est pas en train de charger, le temps de trouver un chargeur et de brancher le téléphone, est juste le temps nécessaire au logiciel pour s'emmêler les pinceaux. Un reboot de la machine est nécessaire après avoir tuer le logiciel de mise à jour Nokia, pour que celui-ci reconnaisse de nouveau le téléphone quand on le branche sur la machine.
Je regrette de devoir faire ce genre de chose sous windows - si Nokia a l'intention de ne pas se laisser marcher dessus par Apple, il va falloir que ce genre de chose soit faisable depuis un Macintosh.
Bon mon firmware est passé de la version 2.x à une version 4.x, quand tout fonctionne cela prend quand même 20 minutes.

Today I once again verified that the world I live in is small very small. I attended Atlassian's user group meeting today, as i use daily both confluence and Jira. It's interesting to meet other users and exchanging on how they use the products. We had some heads up on the upcoming versions - I made a point to tell what I didn't like about both products. I think I would have gotten more about the meeting if I was developing plug-ins or administering the products, but nevertheless I gave a few tips here and there and I really love doing that. Pepijn probably got some information he wanted, and he took note of a plug-in I was interested in. At some point in one of the talks, semapedia was mentioned. When the project was described I immediately recognized semacode, which I know the founder from mid-2003 when I started working on Camino. So some guys from Australia , talk about a Canadian known to a Frenchman that moved to the Netherlands , yep the world is very small.

So I found a bug in one of Hewlett-Packard's printer manuals. I've been poking around HP's website but was not really able to find a simple way to report this issue on page 51 of LJCP3505_use_ENWW.pdf which in the section of Use features in the Macintosh printer driver one can read :

Use automatic duplex printing
1. Insert enough paper into one of the trays to accommodate the print job. If you are loading special
paper such as letterhead, load it in one of the following ways:
● For tray 1, load the letterhead paper face-up with the bottom edge feeding into the printer first.
● For all other trays, load the letterhead paper face-down with the top edge at the back of the
tray.
2. Open the printer driver (see Change printer-driver settings for Windows on page 38).
3. Open the Layout pop-up menu.
4. Next to Two Sided Printing, select either Long-Edge Binding or Short-Edge Binding.
5. Click Print.
Print on both sides manually
1. Insert enough paper into one of the trays to accommodate the print job. If you are loading special
paper such as letterhead, load it in one of the following ways:
● For tray 1, load the letterhead paper face-up with the bottom edge feeding into the printer first.
● For all other trays, load the letterhead paper face-down with the top edge at the back of the
tray.
CAUTION: To avoid jams, do not load paper that is heavier than 105 g/m2 (28-lb bond).
2. Open the printer driver (see Change printer-driver settings for Windows on page 38).

As one can see through my emphasis a bad copy/paste was done from the windows section to the mac section. So that's really minor, but I would like to report it so HP can update it's manual when they decide to release a new version of the drivers.

Je viens de changer d'hébergeur - j'ai quitté à regret nFrance conseil, pour revenir chez Free. Les raisons de ce changement - du hardware veillissant, une offre que j'utilisais et qui n'était plus au catalogue, et enfin un meilleur prix et une offre de service plus importante chez Free.

So I'm about to add some content to flickr, and I want that content to be found. So the idea is to add tags as these will serves as keywords for searches queries. The issue I have with tags is when the tag itself is build with more than one word. Say for instance "Golden Circle" will be the tag for a few pictures I have taken. My issue is with the way I can add a multi-worded tag, I can add "Circle" + "Golden", or "Golden Circle" or the three of them. I think that adding the three of them is not efficient and I like things most of the time to be efficient. The issue is that flikr will change "Golden Circle" to "GoldenCircle". Anybody as hints on how tags should be used in this case ?

At least they are hiring people to at maintain the groups. I just love using mailing lists and usenet over web based forums.

I just tried to synch calendars using their tools to my ical installation and I lost most of it. Zimbra you just suck.

Yesterday I attended the Wireless Community Camp 2007 organized by http://www.wifisoft.org, Held in Oegstgeest the Netherlands. I attended because I was curious on what two of my colleagues where up to - while they did some infrastructure work in June. The Camp Lasted one week, but I only attended Saturday. So I arrived late and came in the middle of a presentation of FON.
Fon is a project aimed at sharing your bandwith with others over a WIFI network. The benefit you get by doing so is that you can use all other FON hotspot around the world for free. If you are not a member of fon you can become a member for a small fee. The idea is to broadcast two networks from the same Wifi Access point, one is going to be the FON network, the other one which is open (but supposed to be protected by the first one), is your personal network.
I was temped to join but as I live on the 6th floor people in the street would probably not benefit from my FON access point. Then I discovered that I would probably need a La Fonera device, a small access point. While the device looks nice and cool I have two issues with it. I usually like to be able to plug cables in my routers not just Wifi ones, and on that thing I just can't. It heats to much. But nevertheless the project seems interesting and I was able to spot two Fonspots, one in the apartment building I live in den Haag, the other one near my parents.
There is also a bunch of software being develop to turn your mac or your PC into a FON access point, the Mac version does not on PowerPC :-(.
Then the camp headed two the center of Leiden to take a boat tour. The idea was to detect as many WIFI signal as possible. The boat trip was fun and I discovered parts of Leiden I did not know. I did not participate to the detection contest.But while we where cruising the canal that split the central market in two, the guy looking for discoverable Bluethooth device food 300 of those !! Then we went back to the camp to get a nice barbecue. I really enjoyed my day and will probably spend more time to WCC next year.

ni les sparcs - mais ça me rappelle 1995. Mais le mail de Theo est très intéressant à lire.

Hier en regardant quelques video dont une vidéo sur l'histoire de la légion étrangère, J'ai eut le besoin de situer l'Angola sur une carte de l'Afrique. J'avais au préalable mis comme moteur de recherche par défaut Yahoo!, le premier résultat est une société sud-africaine qui édite des cartes d'Afrique, mais qui n'en affiche pas sur son site web. La même recherche sur google me donne des carte africaines sans que j'ai besoin de cliquer sur un lien pour y accéder. Si je me rend sur le site de Yahoo! France, les résultats sont légèrement plus pertinent mais pas beaucoup mieux. Je teste régulièrement Yahoo, particulièrement lorsque je suis insatisfait d'une recherche avec google - les résultats sont parfois identiques, souvent différents, les résultats de google étant généralement plus précis. Une des grandes force de Yahoo était son annuaire mais celui-ci ne semble plus être maintenu - c'est dommage. Idem pour l'ensemble des listes de discussions qui sont trés rarement retourné par le moteur de la même maison.

Flickr a récemment mis en place une carte interactive qui permet de geotagger ses photos. Je trouve cet outils très pratique. Mes photos sont donc presque toutes geotagged

Today and yesterday I attended Apachecon europe 2007, in Amsterdam. Previous conferences of such type I attended included : Be's European developers conference, Geektea's at beeurope's headquaters and a few FOSDEM (2004, 2005, 2006 and 2007). I wanted to go to Apachecon for two reasons, the first and foremost is the fact that many of my collegues are apache members, or affiliated with some kind of Apache project. Coming from the client side of things (Camino, mozilla), I wanted to see how things where organized on the other end of the web software land. I've never been to mozilla held conferences so I can't really compare. The entrance price tag to the Amsterdam conference was an issue for me - 500 € per day is a bit excessive from my used perspectives of Free conferences.

The first day I arrived early, which was good because it gave me the opportunity to get my T-shirt, and then connect to the Wifi network and start reading my emails and finish releasing our beta for friends release. I then attended a few tracks :
Lucene
Explained what it is, gave a few examples on how to use it and who uses it (technorati wikipedia).
Presentation is available here.
Lucene Advanced
Gave a few hints on getting better results and performance out of Lucene's indexes. And out to configure it. Specifically to have a look into contrib/benchmark. Better indexes can be built with the stopword option : it removes a word from having weight in the index process so if all the document you index talk about say HTML - it might be a good idea to use the stopword onHTML.
Semantic web without calling it semantic
A Nice presentation about what actually RDF is, what it should be used for, and some of the tools, available to manipulate RDF (things like ANT. Then I followed a presentation on ivy which is a dependency manager that work pretty well with ANT. Key thing to remember from this is that ANT+ivy should be compared to Maven 2, and not maven against ivy alone.
Then back to do or try to do some work - now that we are live we still need to test a few things.
Then Came the presentation on hadoop - which is used at Yahoo!, but not on production systems.
I have issues with the fact that hdfs master nodes do not do any kind of fail over - so if you are not careful and that node crashes, you've lost your file system. The presentation was oriented more torwards the map/reduce computations than on the hdfs - which I would have interest in.
But the Question and Answer part of that talk was very very interesting.
The last thing I assisted was a presentation of Apache Directory Server, which is a "new" OpenLDAP implementation. This brings competition to current LDAP implementation and that is always a good thing. The presentation was done by the project leader so it was fresh and up to date.

Lightning talks is a thing not to miss. And I had the pleasure to see a living legend : Wilfredo Sanchez.

La boite dans laquelle travaille mon cousin a enfin mis du contenu sur son site web. Je sais maintenant ce qu'ils font. L'idée n'a rien de révolutionnaire - mais je pense que le concept est bon et qu'il peut être porteur. Il s'agit d'avoir un numéro de téléphone unique, qui re-dirige les appels vers son mobile, son fixe, celui du bureau. Le bon plan dans cette affaire est que l'on garde le même numéro de téléphone - même en changeant de job , ou bien d'opérateur.

Software patent in Europe

An interesting talk where a nice effort was announced ethipat.org for an ethical patent system. Which want to make sure that in the EU we get a decent system patenting system.

One laptop by child

Very interesting presentation, with new wireless network protocol - the problems encountered are around energy consumption - and we where explained how the screen works which reminded me how the memory mapping for the graphics was done on the Apple II. The project takes everything into account not just software hardware issue. Very very interesting project - oplc.org

Liberating Java

Java should be completed released under GPL v3 by spring. They will use Mercurial as SCM. You will need a few binary plugs - because some of the code is licensed to third party and SUN does not own the rights to release it.
And if some people think sun is screwing up things, Fire an email to simon.phipps@sun.com, they prefer to resolve the issue than having people complaining on webpages.

Here is the repartition of users attending the GnuPG signing key party at FOSDEM 2007. This is all but scientific - because for instance CH has one entry but I know for sure two person will be coming from switzerland - it's based on the TLDs of the emails used for PGP.

As some of you may know I work in the QA department at the Venice Project. Up until recently I was the QA department by myself. This is no longer true. We do all our testing manually which is bad - because humans do fail - whereas software is predictable and will always fails the same way.
Up until now we are using Confluence to track our work and our test cases and testings. But this has limits. I would like to add steps to our buildbots build processes to do unit testing and keep track of those - and I would love to keep track of the memory and disk footprint used by our application. There are a few plug-in for confluence that I could use, but they are more oriented in a manual reporting - people click to say if test was successful or not - this is not really compatible with the use of bots to do unit testing. I've done my bits of googling and found many pages describing software that might achieve what I want. I also asked my question on usenet - and up to now had one answer giving me the name of the type of software I'm looking for and a link to a german study/comparison of those pieces of software. Still I would love to get live feedback about the usage of such software so I could make a choice. My needs are :

• Document test cases.
• Write test plans.
• Track test plans - manual testing - .
• Get input data from external source to track what my bots are doing.
• Be Able to track performance for clients (memory, disk footprint).
• Be Able to track performance for Server applications (Stress testing, memory usage, cpu usage)

I need to be able to extract reports on a weekly basis, and do some stats on at least an annual basis (choosing from month y to month x - see the number of test runs, with failure/pass, memory usage changes over a long period etc ...).

Integration with svn, Jira are pluses.

Voila - comme je ne peux pas accèder au web avec mon nokia N73. Je l'utilise comme un appareil photo, et j'ai décidé d'y installer quelques jeux. Enfin c'est un pluriel un peut rapide, pour l'instant je n'ai qu'un seul jeux, il s'agit de Lucky Luke.
La musique est simpliste mais elle ne couvre pas les bruitage du jeux et dans ce genre shoot'em up, je pense que c'est important. Le jeux est disponible en français, et en presque toutes les langues européennes. La prise en main du jeux est assez facile - mais l'aide en ligne n'est pas complète car elle n'indique pas que le joystick de mon nokia est pris en compte, il m'a fallu trois bonne partie pour m'en rendre compte. Et c'est nettement plus agréable de jouer avec le joystick qu'avec le clavier du téléphone. J'ai juste un peut peur de le casser - et donc j'ai arrêter de jouer avec pour cette simple raison (par arrêter de jouer j'entends toutes les 5 minnutes). Un jeux simple mais efficace - et qui plus est est une réalisation française du studio