Tuesday, September 30, 2008

Java Bar Camp Mardi 30 septembre

J'ai participé hiers soir à mon premier "Java Bar Camp" à Paris, dans les locaux de Octo Technologie.

Avant d'aller plus loin, je tiens à remercier les organisateurs, mis à part le retard sur l'arrivée des participants, tout le reste a été très bien géré, bravo !

Premières impressions : ce n'est pas du tout ce à quoi je m'attendais. Je m'attendais à un rassemblement de geek & développeurs barbus, alors que l'assemblée était beaucoup plus hétéroclite, avec une forte coloration consultant/chef de projet.
Les thèmes s'en sont fortement ressenti : méthodes de développement, build et tests étaient à l'honneur.

Mais revenons sur le déroulement de la soirée :
- 18h30 / 19h30 arrivée des participants (un peu de retard, Julien Dubois (M. Spring en France) absent (il est papa :), toutes mes félicitations )
- 19h30 :
* tour de salle : chacun décline nom/prénom, et 3 tags, des sujets qu'il souhaite évoquer - dans un bar camp, tout le monde est acteur ;
* après le tour de salle, chacun est libre d'aller inscrire sur un post-it le sujet dont il souhaite parler parmi les tags;
* 8 sujets sont choisis parmi les post-it, organisés en 2 sessions (donc réparti dans 4 salles)
- 19h45/20h35 : première session : TDD/Test ; Licenses OSS ; Spring vs JEE ; ? ;
- 20h40/21h30 : seconde session : usine logicielle (build, maven...) ; scrum en vrai ; ? ; ?
- ensuite, buffet et discussion libres.

J'ai choisi en première session JEE5 vs Spring (à regret), et en seconde session "Scrum en vrai".

Je ne vais pas m'éterniser sur la première session, orienté "débat", et il fut relativement stérile et consensuel. Rien n'émerge clairement, sauf que "certain client ne veulent que des normes, et JEE s'impose", "Sans Spring, Java en entreprise serait sûrement mort, bouffi par sa complexité". Bref, rien de nouveau. Le point de vu de Julien aurait sûrement mis un peu de couleurs.

La seconde session m'a beaucoup plus plû. J'y ai rencontré Eric Mignot de Pyxis, un "évangéliste Scrum", intégriste (dans le bon sens ), qui m'a permis de voir le chemin qu'il reste à parcourir dans sa mise en place dans les projets auxquels je participe, mais aussi de prendre consience du chemin déjà parcouru.

Après avoir revu les bases de scrum (organisation, objets), nous avons abordé les problèmes contractuel, la gestion de l'architecture, la gestion humaine des équipe.

Le mot qui revient le plus souvent est "engagement" : un projet Scrum ne peut pas fonctionner si toutes les personnes y prennant part son convaincu par la méthode, en particulier si le scrum master n'a pas un soutient indéfible de sa direction, et si le product owner n'a pas le pouvoir décisionnaire sur les paiement/le budjet et sur le périmètre fonctionnel.

La fin de la soirée étaient agréable, les groupes se formant et se séparant au gré des discussions, souvent pationnée.

En résumé : une expérience intéressante, un fin de soirée nettement plus intéressante que le début, et je reste sur ma fin en terme de développement pur - ce n'est pas dans ce genre de soirée que je parlerais de Scala ou de language fonctionnels :)

Project Euler - Scala

I discovered Project Euler, a website that purpose a lots of Mathematical problems that can be solved with programming. The website provides a check of you solution, and a forum to discuss it but only after you give the rigth answer.

Scala really shine in this kind of game. Well, I think that any functionnal language, and perhaps Ocaml even better that other (thanks to it's amazing vm), are really well fitted for this kind of small, algoritm instensive tasks.

The first problems are really simple, and brute force works well, but quickly it is not enonght, and the game become really interesting : it force myself to remind old memories of school, and I'm afraid at the size of the hole in them - I mean, I should have followed some advandced Maths courses in classe préparatoires, and it's almost if I have to discover again all the concept and properties and theorems... A good reason to do all the Project Euler problem till the last one !

Amazing Mythical Man Month

Fred Brooks was rigtht. Well, I know I will not have a Nobel for this fantastic discovery, but each time I refer to the TMMM, I'm really impressed by the insight of the man.

Ok, so, to be more specific : TMMM has 33 years. He is older than me. And Fred Brooks already depected most of the Agile programming method (history is constantly repeating itself).

But perhaps what impresse me the most is that he forecasted 33 years ago what is happening with Open Source software and the JVM ecosystem today : we will build better and cheaper programms only if common, higth-quality tools and library are broadly available.

It's (almost, but if we look from a hight level, it's not too bad) what is happeing in the JVM (do not read Java) world, and tools like IVY and Maven,
Superpackages (ah, not yet) only make it happens quicker.

The future will be intersting, if Sun does not break everything (I hope it's too late for that now).

Monday, September 29, 2008

About me

OK, since I broke the rule I decreed in the first post, at least I should let some information about me.

I'm François Armand, I'm French (Paris) and work in software development for an open-source company, Linagora. My daily work is mostly with Java, with all the drowback it brings - and some quite good surprises from time to time, especially about the wealth and diversity of its ecosystem (but this point is more about the JVM's ecosystem than Java-the-language).

I'm rather new in computer related stuff, since my first programms where done when I reach the master cycle (don't now exactly how French's 'Grandes Écoles" translate in other cultures - for whom who migth be curious, I'm from "ENSIIE"). It was in september 2001.

I keep from that period a deep interest for functionnal programming, and especially a real good memory in my 2004 research project (the one for the end of school's master cycle), when I was allowed to take part of the INRIA's CompCert (Certified Compiler) project and had a deeper look into Ocaml and COQ (a great program prover - well, not exactly, see by youserlves). I will thanks for that Sandrine Blazy and all the people I met in this time (Xavier Leroy, Yves Bertot and others).

The following is more boring. I met the "real world, with real programs and stuff that matter", say Java and Web development. Hum. Well, all is not dark in this side of the world, because I work in company deeply involved with open source software, and I take part of some really good projects, as for example the FederId project, in which I was the developer of InterLDAP-WUI.

All in all, I'm not a real common Java developer, and it's perhaps why I'm looking toward Scala, or Tapestry 5.

OK, that all for now, just a last sentence to link my linkedin profil (I'm also on viadeo, and I believe that the two are rather out of date).

Tapestry 5, Scala

Since the first time I played with Tapestry 5 about 18 months ago, I know it would be great, and the more I use it, the more I see how well thought it is. But again, the more I see the limitation of Java-the-language.

I really need to test again Tapestry 5 with Scala as a back-end - Yeah, Scala is the other hot-thing I follow since about the same time, and at that moment Scala/T5 was working fine. But hence Scala evolved a lot, and T5 relies deeply on Javassist and byte code manipulation, I fear that I may encounter some bad surprises, especially in real-life application.

One more item in my to-do list :)

Tuesday, September 23, 2008

Java 7, aka the Rubicon

I'm looking forward to write something about Java 7, and why it makes me feeling more and more uneasy. To summarize my thinking, I believe that Java, the language, can not continue to evolve without a clean deletion of old, not alway smart or still relevant design choice of java 1 to 5, but that the JVM have a great deal to play as the new open runtime environment.

Hope I will find some time to develop it a little, at least it's writing somewhere.

Tuesday, September 16, 2008

Hard to get OpenId

The so called standard authentication standard, with all its hole and drawbacks, is becoming a required tools... So, I have to open a blog to get an OpenID (google already has so many infos about me... Of course I trusted them, they are so nice...).

Don't expect real post here :)

  © Blogger template 'Minimalist G' by Ourblogtemplates.com 2008

Back to TOP