Le blog d'Arhuman

Thanks to the CPANtesters Wiki , It was quite straightforward to setup my Windows testing box. Typing ' cpan install CPAN::Reporter ' was all needed, installing the dependencies along the way. As I planned to do some smoking tests, I also had to install CPAN::Reporter::Smoker : ' cpan install CPAN::Reporter::Smoker ' Some modules don't play well with (automatic) testing, so configuring distroprefs is probably a good idea : After entering "C:\Documents and settings\UserName\Mes Documents\" you only have to create .cpan/prefs to put there the YAML files downloaded from   DAGOLDEN's site Then typing ' o conf init test_report ' inside the cpan shell and answering the few configuration questions should complete the process.   Now every module installed (or just tested) will produce a report that you can send (or will be automatically sent according to your setting) and entering a smoke test is just a matter of typing : perl -MCPAN::Repo...

My initial intent was too install VmWare Server (which is free (as beer), and that I used a long time ago with great satisfaction) on my Ubuntu boxes. But as it appeared that my kernel (and all kernels 2.6.30 and above) now requires an unofficial additional patch to fix the VmWare Server installer * , I decided to play it safe (or at least simple) and install the VmWare Player 3.1.0 instead. * See http://communities.vmware.com/thread/215985 for more information The VmWare Player install went fine, and installing an XP virtual machine was just a matter of minutes. For some reason NAT was not working out of the box, so I only changed the VM network setting to Bridge and was then able to download Strawberry Perl 5.12.0.1. Sadly my attempt to install also release 5.10.1.2 in the same VM failed miserably (the installer detecting an higher version)... If anyone managed to get (simply) 2 version on the same Windows box, I'd be glad to hear how... Now I still have to install...

CPAN Smoke testing is basically, executing automatically unknown code at a large scale on a machine you own. And if this risk is negligible (to me) for well known modules, it has to be reevaluated if you plan, as I do, to test numerous unknown modules.  That's why my initial plan to do some automatized tests with unprivileged user is going to be slightly modified with the addition of virtualization. In fact, the more I think about it, the more I wonder why nothing really bad has already happened with so many scripts blindly executed all around the world... Anyway, I'm currently setting a VmWare infrastructure and I hope to be able to give you some results soon... [French Translation] Les smoking tests CPAN sont à la base l'exécution automatique de code inconnu à grande échelle sur une machine que l'on possède. Et si le risque est négligeable (pour moi) pour des modules bien connus, il doit être réévalué si vous prévoyez, comme je l'envisage, de tester de ...

I've been a huge fan of CPAN Testing for quite a long time now. I've even been a CPANTester. But things have changed a lot, the tools are now numerous and impressive to allow bigger and better testing. Some testers achieved more than 70 000 reports a month ! As a stat junkie I see it as a challenge... My first step to become a CPANTester is to allow sending report from my home (dynamic ADSL IP) server, through my online hosted virtual server. As it took me some time to setup the proper configuration * , below are the postfix configurations for SASL authentication on the SMTP server and client : * So much time in fact that, I even thought about ssh tunneling [French translation] Je suis un grand fan du testing CPAN depuis un bon moment. J'ai même été un CPANTester autrefois. Mais les choses ont beaucoup changé, les outils sont maintenant nombreux et impressionnant pour permettre de tester plus et mieux. Certains testeurs arrivent à produire plus de 70 000 r...

I came across a comic that illustrates well, to my mind, the issue with Perl's popularity : XKCD : The secret of the Universe Wherever the place I've worked, Perl was present to one degree or another, often in a crucial place, without its use being advertised or even known. I've long believed, that it was the reason why Perl wasn't as popular as it could be, despite its use. But in fact, no : It's just that the Perl community sucks at marketing ! It's the community duty to promote the language, to bust the myths, to give visibility to the language, for we can see one day more jobs and more formations mentioning explicitly the language. That's why initiatives around Enlightened Perl must be supported : The Iron Man project especially : To post regularly about Perl to give it visibility. (and by the way to achieve the honorific Iron Man status without running 42km, swimming 5km and riding 180km on a bike ;-) ) Then, here it is, by this post...

J'ai longtemps pensé que j'aimais bien améliorer la roue : prendre les choses et les pousser un peu plus loin, améliorer ce qui pouvait l'être, corriger ce qui n'allait pas, trouver une nouvelle utilisation, etc... Mais en y réfléchissant ce n'est pas tant les choses que la démarche qui m'importe : C'est plutôt "continuer à faire tourner la roue de l'amélioration"... Pour moi améliorer un objet, optimiser son code, s'affuter physiquement, mieux connaître les autre, c'est la même démarche. Mais peu de personnes l'appliquent à tous les domaines. Je suis informaticien, mais je trouve dommage de n'être qu'un "Pragmatic programmer" alors qu'on peut être un "life hacker" (sans les majuscules sinon il y a un trademark...) Alors voilà, aujourd'hui je ne parlerai pas d'outils, mais de démarche. Et tant que j'y suis je vous livrerai "ma méthode personnelle à moi que je l'ai faite tout seul...