01 February 2020

« Gagner la guerre » de Jean-Philippe Jaworski

Grand amateur de fantasy, la plupart de mes lectures sont généralement en anglais. « Gagner la guerre » de Jean-Philippe Jaworski trônait dans ma pile (virtuelle) de lecture depuis un moment, auréolé par ses excellentes critiques et la perspective de lire une œuvre de fantasy en français. Il y était accompagné de « Janua Vera », le recueil de nouvelles dans le même univers précédent le roman.

Jaworski n'a pas usurpé sa réputation d'excellent auteur, son texte est incroyablement bien écrit. Le style et le rythme sont remarquables et je me suis plusieurs fois surpris à relire certains passages rien que pour le plaisir d'en profiter une seconde fois. Le vocabulaire est également extrêmement riche; c'est bien simple, j'ai dû plus souvent utiliser la fonction dictionnaire de ma liseuse que lors de mes lectures en anglais !

L'univers, si bien introduit dans « Janua Vera », est toujours aussi cohérent et plaisant à découvrir. Ce mélange de pseudo-réalisme historique, la ville principale savant mélange de Florence et de Rome antique, saupoudré de magie et de fantasy fonctionne à merveille.

Toutes ces belles qualités sont malheureusement ternies par un propos extrêmement viriliste et des personnages féminins quasi inexistants. On retombe de plain pieds dans les critiques et clichés souvent associés au genre et c'est bien dommage. Cela m'a d'autant plus marqué après les nombreuses œuvres de Brandon Sanderson et Robin Hobb que j'ai lues récemment et qui ont démontré avec brio qu'on pouvait écrire de l'excellente fantasy avec des personnages féminins forts et intéressants. J'ai un peu le même arrière-goût qu'après la lecture de « La Horde du Contrevent » qui tombait dans les mêmes travers, bien que de façon moins marquée. Cela donne l'impression que la fantasy française est restée bloquée au siècle passé et n'arrive pas à sortir des stéréotypes de genre qui ont trop longtemps collés à ce style littéraire.

Du coup si vous avez des recommandations d'auteurs·rices francophones qui arrivent à éviter ces écueils je suis preneur.

30 January 2020

Dealing with Loss

Warning: This blog post contains a lot of talk about feelings, loss, and discussion of a suicide.

Recently, I have been thinking a lot about loss. My nephew died just a few months ago, after a short life with Duchennes Muscular Dystrophy. A neighbour recently took her own life, leaving a husband and two children behind. And today I learned that someone I have known for 15 years in the open source world recently passed away through a mailing list post. In each case, I have struggled with how to grieve.

My nephew has been ill for a long time, and we have been open in my family about taking advantage of opportunities we have to spend time with him for the past few years, because we knew he would not live much longer. And yet, death is always a surprise, and when we got a phone call one Saturday in November to let us know that he had passed away in his sleep, my first instincts were logistical. “I have a work trip coming up – when will the funeral service happen? Can I travel to Asia and get home in time, or do I need to cancel my trip? What is the cheapest way to get home? Who should travel with me?” When I got home, the funeral is a multi-day collective grieving, with neighbours, cousins, uncles and aunts arriving to pay their respects, express their condolences, spend time with the family. It was not until we were shovelling dirt on top of the casket that I really thought about the finality of the burial – I will never see my nephew again.

And yet, I was not overwhelmed with grief. I have never really known him intimately. How well do you know a child 25 years your younger, after you leave home and live abroad? How close of a connection do any thirty-somethings have with their teenage nieces and nephews? I second-guessed my emotions. Should I feel sadder? Is there an appropriate way to grieve? In the end, I decided to allow myself to feel the feelings I felt, and not to try to figure out whether I “should” be feeling differently. But avoiding self-judgement was difficult.

Last week, when we got the news about our neighbour, it hit me pretty hard. We knew the family well, had been to barbecues and play-off games in their house. I had coached basketball with her husband, one of their sons was in the team. Initially, we read that she had “passed away suddenly”, it was only through school bus stop gossip that we learned that she had committed suicide. We learned that she had been suffering from depression, that her life had not been easy for the past few months. I felt a great sadness, and also a little guilt. We had enjoyed her company in the past, but I knew nothing of her life. I was about to leave on a work trip, I would miss her memorial service and funeral. I was told that the ceremonies were very emotional, and really felt like the community coming together. The priest leading the service spoke openly about suicide and depression, and my wife said that his ceremony gave her a great sense of peace, removing the veil from some of the awkwardness that she felt around the topic. It gave the community an opportunity to start healing.

But I was not there. Now, I have all of these other thoughts about the appropriate way for me to grieve again. My instinct is to call to their house to express my condolences, but I am afraid to. This time, I find myself comparing my feelings to those of her family. I imagine how they must be feeling. Surely they are devastated, probably angry, maybe even feeling guilty. I think about her sons, the same age as two of my own sons, and I wonder what their lives will be like now. What right do I have to feel grief, or to impose on their grieving to express my feelings to them? How would I react, in the same circumstances, if this acquaintance called to the house a week after a funeral ceremony? And then, I also feel guilt. Sure, we didn’t know each other that well, but could I have been there for her in some way? Was there some way that we could have helped? I think about how alone she must have felt.

And now, today, I have learned of the death of someone I would have called a friend. Someone I would regularly meet at conferences, who I got along very well with professionally and personally, two or three times a year. I was not a part of his life, nor he a part of mine. I’ve found myself tearing up this morning thinking about our interactions, realizing that we will never meet again. And once more, I struggle to find the appropriate way to grieve.

I don’t know why I felt compelled to write this – I have debated saving it as a draft, deleting it, writing it in a private text file. But I am sharing it. I think I feel like I missed a part of my education in dealing with loss. I feel like many people missed that part of our education. Maybe by sharing, other people can share their feelings in comments and help me further my own education. Maybe by reading, others who struggle with dealing with loss will realise they’re not alone. Maybe it will achieve nothing more than helping me deal with my own feelings by verbalizing them. Let’s find out…

12 January 2020

Buster, jack, systemd, etc.

Nouvelle année et dans la foulée d’une série de mises à jour relatives à Python 3, quelques jours off ont aussi été l’occasion de reprendre quelque peu certains systèmes de la radio. D’abord des trucs évidents comme des mises à jour en retard, ensuite des modifications au démarrage des services audio, pour profiter de systemd.

Jusqu’ici ils étaient gérés par supervisor mais je n’ai jamais été grand fan, et avec systemd il y a désormais tout le nécessaire pour gérer de manière native ces services. J’écris « ces services » mais quels sont-ils ? Il s’agit de jack, infra générale du système audio, jack-plumbing, bricoles pour assurer le branchement entre éléments audio, liquidsoap, enregistrement (pige d’antenne) et streaming, et openob, lien basse latence vers l’émetteur FM. Tout ça se retrouve composé dans différentes machines virtuelles, pige, streams, openob, qui tourne sur une machine hôte avec la carte son, la liaison audio entre l’hôte et les machines virtuelles se fait alors en utilisant le backend réseau de jack, sur une interface multicast.

Le paysage posé, démarrage de la mise à jour système vers Buster, et première surprise avec le multicast, justement, qui ne semble plus fonctionner. Il y a eu du neuf côté smcroute, et le script wrapper devant assurer une compatibilité semble échouer sur la configuration pourtant basique, qu’à cela ne tienne, il y a désormais un fichier de configuration et tout simplement y inscrire mon souhait, mgroup from eth0 group, remet le tout en place.

La suite c’est la création des unités systemd, jackd.service, jack-plumbing.service, openob.service, liquidsoap.service, qui n’ont absolument rien de particulier, juste une attention portée à l’ordre de démarrage, un peu sensible (et il y aurait sans doute à améliorer un peu les choses avec un sd_notify dans jackd.

Et le résultat, c’est un contrôle avec systemctl et cie, pour par exemple avoir vue générale de l’état par systemctl status ou une vue précise des logs via journalctl, ce qui est un conforme énorme.

│ └─358 /usr/bin/python3 /home/panik/openob/bin/openob brusinik2 [...]
│ └─348 /usr/sbin/smcrouted -n -s
│ └─351 /usr/bin/jackd -r -d net
│ └─361 /usr/bin/jack-plumbing

Reste à faire la même chose à quelques endroits encore, pour l’enregistrement direct des studios et sur l’ordinateur posé côté émetteur FM mais il ne devrait y avoir là aucune surprises (...), ça peut attendre d’autres congés.

10 January 2020

Rust/GStreamer paid internship at Collabora

Collabora is offering various paid internship positions for 2020. We have a nice range of very cool projects involving kernel work, Panfrost, Monado, etc.

I'll be mentoring a GStreamer project aiming to write a Chromecast sink element in Rust. It would be a great addition to GStreamer and would give the student a chance to learn about our favorite multimedia framework but also about bindings between C GObject code and Rust.

So if you're interested don't hesitate to apply or contact me if you have any question.

30 December 2019

Deux ans et demi

Ça commençait vraiment à faire longtemps, espace délaissé, heureusement tic toc du calendrier de fin de vie de Python 2, l’occasion de reprendre ici, réécrire de manière minimale toute l’affaire (code), juste à temps, importer les vieilles notes, en relire pas mal, en redécouvrir certaines, ne pas toutes les republier.

Esperanzah! 2017 c'était le dernier post, il y eu 2018, il y a eu 2019, se prépare déjà 2020; depuis 2017 il y a bien sûr Panik, avec une place que les archives passées n’exposaient pas tellement, avec une matinale qui continue à aller fort, avec des PBG où j’ai pris une petite place — et je devrais parfois profiter de ces émissions/chroniques pour poster ici, avec Alerte Niveau 5 qui a repris en novembre du poil de la bête, et le reste.

Côté logiciels libres c’est devenu léger pour GNOME, on est même parvenu à ne pas faire de release party à Bruxelles pour la 3.32, côté Debian c’était déjà léger, à part ça quelques contributions éparses, mais c’est surtout dans les projets boulot (et radio) que le code se fait.

Pour 2020 on verra donc si j’écris davantage, qui sait, cet espace deviendra peut-être même un blog à chats,


Le chat de la voisine, 30 décembre 2019

17 December 2019

GMemoryMonitor (low-memory-monitor, 2nd phase)


Use GMemoryMonitor in glib 2.63.3 and newer in your applications to lower overall memory usage, and detect low memory conditions.


To start with, let's come back to low-memory-monitor, announced at the end of August.

It's not really a “low memory monitor”. I know, the name is deceiving, but it actually monitors memory pressure stalls, and how hard it is for the kernel to allocate memory when applications need it. The longer it takes to allocate memory, the longer the kernel takes to allocate it, usually because it needs to move memory around to make room for a big allocation, when an application starts up for example, or prepares an in-memory buffer for saving.

It is not a daemon that will kill programs on low memory. It's not a user-space out-of-memory killer, and does not take those policy decisions. It can however be configured to ask the kernel to do that. The kernel doesn't really know what it's doing though, and user-space isn't helping either, so best disable that for now...

As listed in low-memory-monitor's README (and in the announcement post), there were a number of similar projects around, but none that would offer everything we needed, eg.:
  • Has a D-Bus interface to propagate low memory conditions
  • Requires Linux 5.2's kernel memory pressure stalls information (Android's lowmemorykiller daemon has loads of code to get the same information from the kernel for older versions, and it really is quite a lot of code)
  • Written in a compiled language to save on startup/memory usage costs (around 500 lines of C code, as counted by sloccount)
  • Built-in policy, based upon values used in Android and Endless OS

Next up, in our effort to limit memory usage, we'll need some help from applications. That's where GMemoryMonitor comes in. It's simple enough, listen to the low-memory-warning signal and free some image thumbnails, index caches, or dump some data to disk, when you receive a signal.

The signal also gives you a “warning level”, with 255 being when low-memory-monitor would trigger the kernel's OOM killer, and lower values different levels of “try to be a good citizen”.

The more astute amongst you will have noticed that low-memory-monitor runs as root, on the system bus, and wonder how those new fangled (5 years old today!) sandboxed applications would receive those signals. Fear not! Support for a portal version of GMemoryMonitor landed in xdg-desktop-portal on the same day as in glib. Everything tied together with installed tests that use the real xdg-desktop-portal to test the portal and unsandboxed versions.

How about an OOM killer?

By using memory pressure stall information, we receive information about the state of the kernel before getting into swapping that'd cause the machine to become unusable. This also means that, as our threshold for keeping everything ticking is low, if we were to kill high memory consumers, we'd get a butter smooth desktop, but, based on my personal experience, your browser and your mail client would take it in turns disappearing from your desktop in a way that you wouldn't even notice.

We'll definitely need to think about our next step in application state management, and changing our running applications paradigm.

Distributions should definitely disable the OOM killer for now, and possibly try their hands at upstream some systemd OOMPolicy and OOMScoreAdjust options for system daemons.


Creating low-memory-monitor was easy enough, getting everything else in place was decidedly more complicated. In addition to requiring changes to glib, xdg-desktop-portal and python-dbusmock, it also required a lot of work on the glib CI to save me from having to write integration tests in C that would have required a lot of scaffolding. So thanks to all involved in particular Philip Withnall for his patience reviewing my changes.

13 December 2019

Dual-GPU support follow-up: NVIDIA driver support

If you remember, back in 2016, I did the work to get a “Launch on Discrete GPU” menu item added to application in gnome-shell.

This cycle I worked on adding support for the NVIDIA proprietary driver, so that the menu item shows up, and the right environment variables are used to launch applications on that device.

Tested with another unsupported device...

Behind the scenes

There were a number of problems with the old detection code in switcheroo-control:
- it required the graphics card to use vga_switcheroo in the kernel, which the NVIDIA driver didn't do
- it could support more than 2 GPUs
- and it didn't really actually know which GPU was going to be the “main” one

And, on top of all that, gnome-shell expected the Mesa OpenGL stack to be used, so it only knew the right environment variables to do that, and only for one secondary GPU.

So we've extended switcheroo-control and its API to do all this.

(As a side note, commenters asked me about the KDE support, and how it would integrate, and it turns out that KDE's code just checks for the presence of a file in /sys, which is only present when vga_switcheroo is used. So I would encourage KDE to adopt the switcheroo-control D-Bus API for this)


All this will be available in Fedora 32, using GNOME 3.36 and switcheroo-control 2.0. We might backport this to Fedora 31 after it's been tested, and if there is enough interest.

08 August 2019

Ubuntu 18.04.3 LTS is out, including GNOME stable updates and Livepatch desktop integration

Ubuntu 18.04.3 LTS has just been released. As usual with LTS point releases, the main changes are a refreshed hardware enablement stack (newer versions of the kernel, xorg & drivers) and a number of bug and security fixes.

For the Desktop, newer stable versions of GNOME components have been included as well as a new feature: Livepatch desktop integration.

For those who aren’t familiar, Livepatch is a service which applies critical kernel patches without rebooting. The service is available as part of an Ubuntu Advantage subscriptions but also made available for free to Ubuntu users (up to 3 machines).  Fixes are downloaded and applied to your machine automatically to help reduce downtime and keep your Ubuntu LTS systems secure and compliant.  Livepatch is available for your servers and your desktops.

Andrea Azzarone worked on desktop integration for the service and his work finally landed in the 18.04 LTS.

To enabling Livepatch you just need an Ubuntu One account. The set up is part of the first login or can be done later from the corresponding software-properties tab.

Here is a simple walkthrough showing the steps and the result:

The wizard displayed during the first login includes a Livepatch step will help you get signed in to Ubuntu One and enable Livepatch:

Clicking the ‘Set Up’ button invites you to enter you Ubuntu One information (or to create an account) and that’s all that is needed.

The new desktop integration includes an indicator showing the current status and notifications telling when fixes have been applied.

You can also get more details on the corresponding CVEs from the Livepatch configuration UI

You can always hide the indicator using the toggle if you prefer to keep your top panel clean and simple.

Enjoy the increased security in between reboots!




08 July 2019

Bolt 0.8 update

Christian recently released bolt 0.8, which includes IOMMU support. The Ubuntu security team seemed eager to see that new feature available so I took some time this week to do the update.

Since the new version also featured a new bolt-mock utility and installed tests availability. I used the opportunity that I was updating the package to add an autopkgtest based on the new bolt-tests binary, hopefully that will help us making sure our tb3 supports stays solid in the futur ;-)

The update is available in Debian Experimental and Ubuntu Eoan, enjoy!

24 September 2018

Pourquoi les applications Flatpak, c’est l’avenir

Régulièrement, j’entends des utilisateurs se plaindre de ce format de paquet, qui occuperait un peu plus d’espace disque ou qui aurait encore quelques bugs de jeunesse, comme le thème de l’application qui pourrait différer de celui de l’utilisateur ou, plus ennuyeux, l’impossibilité de pouvoir jouir de certaines fonctionnalités, qui seraient pourtant disponibles dans la version standard, non exécutée dans un « bac à sable ».

Mais il faut voir sur le long terme. La version 1.0 est enfin sortie au mois d’août 2018, après plusieurs années de développement, ce qui permet de franchir un premier cap. Maintenant, il faut savoir que si elles n’ont pas été directement pensées pour ce mode de fonctionnement, certaines applications ont besoin d’être adaptées pour pouvoir fonctionner parfaitement. Mais ce n’est qu’une question de temps et ça ne doit pas éclipser pour autant les nombreux avantages déjà permis.

Tout d’abord, nous pouvons citer la sécurité. Les applications Flatpak sont exécutées dans un environnement « bac à sable » (sandbox) sûr, isolé du reste du système. Mieux encore, comme pour les applications mobiles, les développeurs doivent déclarer dans un manifeste de quelles autorisations ils souhaitent pouvoir bénéficier : accès au dossier personnel de l’utilisateur, à certains périphériques (webcam, micro…), à la géolocalisation… Droits que l’utilisateur est libre d’accorder ou révoquer à tout moment. Alors oui, pour un logiciel libre dans lequel l’utilisateur a toute confiance, ça n’a pas forcément grand intérêt, mais dans le cas de logiciels propriétaires, véritables boîtes noires dont on ne sait rien, ça peut tout de suite être plus intéressant.

Autre avantage important, la possibilité offerte aux développeurs de pouvoir atteindre directement l’ensemble de leurs utilisateurs sans attendre le bon vouloir des différentes distributions. Une nouvelle version vient de sortir, un Flatpak est proposé et tout le monde peut en bénéficier, sans avoir à se soucier du système de paquet utilisé par la distribution (DEB, RPM…) ou de la compatibilité des bibliothèques.

Non seulement les développeurs pourront proposer un paquet universel dès la publication de leur application, mais ce dernier pourra représenter la version idéale telle qu’ils l’ont souhaité.

Parce qu’il faut savoir que les paquets des différentes distributions sont souvent bien loin de correspondre à cet idéal. Par exemple, pour des raisons philosophiques ou juridiques, les distributions peuvent très bien désactiver certaines fonctionnalités au moment de la compilation. Des distributions comme Debian ou Fedora, qui font très attention aux quatre libertés du logiciel libre ainsi qu’aux brevets logiciels, préfèrent ainsi se passer de certaines fonctionnalités (par exemple, un algorithme qui serait protégé par un ou plusieurs brevets), plutôt que de se priver de l’application dans son ensemble. Sans parler des nombreux patchs que les distributions peuvent appliquer, dans le but de modifier volontairement le comportement de l’application. Que ces changements soient ou non positifs, l’utilisateur peut très bien préférer la vision des développeurs officiels.

Puis comme les paquets sont identiques pour tous et que les applications sont exécutées dans le même environnement, là encore identique, si l’application fonctionne bien chez le développeur, elle fonctionnera tout aussi bien chez les utilisateurs. Et si l’utilisateur rencontre un bug, ce dernier devrait être plus facilement reproductible par le développeur. Il sera donc bien plus simple d’offrir des garanties et de corriger certains problèmes.

C’est également un gain de temps pour les développeurs, qui n’auront plus à se soucier que d’un unique Flatpak, plutôt que de créer de nombreux paquets pour différentes distributions (quand ils ne se contentent pas, bien souvent, de ne viser qu’une ou deux distributions majeures, laissant les autres sur le carreau).

Alors bien sûr, on se dit que les différentes distributions ont leurs propres contributeurs pour empaqueter toutes ces applications (petite parenthèse pour rappeler que les distributions ne se préoccupent, en général, que de logiciel libre, et que l’éditeur d’une application propriétaire ne bénéficiera pas de toute cette main d’œuvre). Mais quand on y pense, que de temps humain gaspillé à recréer tous ces paquets, chacun dans leur coin, pour les mêmes applications… Sans parler des plus petites distributions, qui n’ont pas forcément les moyens humains de gérer tout ça. Ne serait-il pas plus intéressant de pouvoir créer un paquet universel une fois pour toute, et de pouvoir ensuite se concentrer sur des tâches plus gratifiantes ou plus utiles ?

Autre avantage auquel on ne pense pas forcément, la possibilité d’installer sans risque plusieurs versions en parallèle. Que ce soit une version stable et une version de développement à des fins de test, ou une ancienne version stable qui pourrait proposer des fonctionnalités dont on a besoin mais qui auraient malheureusement été supprimées des versions plus récentes (l’évolution des logiciels que l’on utilise ne nous convient pas toujours).

La compatibilité dans le temps devrait également être renforcée. Aujourd’hui, de souhaiter utiliser de vieilles applications abandonnées par leurs développeurs (et donc non adaptées à des systèmes modernes) peut rapidement devenir compliqué, pour ne pas dire impossible pour la plupart des gens, puisque toutes les distributions n’acceptent pas forcément le risque et la charge de travail supplémentaire que représentent de vieilles applications abandonnées ou des bibliothèques obsolètes. Et si ce n’est pas géré par la distribution ou la communauté, ça implique bien souvent de devoir mettre les mains dans le cambouis et de compiler soi-même. Tandis qu’avec un Flatpak et son runtime d’époque associé (qui contient les différentes bibliothèques nécessaires à son bon fonctionnement), la distribution n’a plus besoin de s’en préoccuper.

Donc même si ça ne vous intéresse pas et que vous ne prévoyez pas de changer vos habitudes, on ne peut honnêtement pas dire que cette technologie n’a aucun intérêt (ne serait-ce que pour tout le temps que ça fait gagner aux développeurs de logiciels libres, qui travaillent bien souvent bénévolement). Tout comme il faut arrêter de penser que les distributions de type rolling release telles que Arch Linux ou Manjaro soient le Saint Graal. La première exclut tous les néophytes et la seconde, qui désactive (à raison) le dépôt communautaire AUR par défaut, n’offre donc pas forcément le même catalogue applicatif ou les mêmes versions. Et bien évidemment, en dehors de leur capacité à proposer des versions plutôt à jour, ces distributions ne règlent aucun des différents problèmes soulevés (sécurité, reproductibilité, compatibilité, gain de temps…).

Il est donc préférable de se mettre un instant à la place de l’utilisateur néophyte qui peut enfin bénéficier, à tout instant et de façon sécurisée, des dernières versions de ses applications préférées sans avoir à se soucier de toutes les questions techniques sous-jacentes, qui ne l’intéressent pas et ne l’intéresseront jamais : le choix de la distribution et du format de paquet qui en découlera, les éventuels dépôts tiers à activer (parfois gérés de façon plutôt hasardeuse, pour ne pas dire risquée), les dépendances nécessaires, l’installation d’éventuels outils de développement pour compiler soi-même et bien faire attention à chaque installation ou mise à jour à ne surtout rien casser…

La question est donc de savoir si l’on souhaite ou non démocratiser l’utilisation de GNU/Linux auprès du grand public. Et si c’est le cas, Flatpak pourrait grandement nous y aider.

07 September 2018

Sortie de GNOME 3.30

Après six mois de développement, comme toujours sans le moindre retard, nous avons droit à une nouvelle version de notre environnement préféré (si si :)

Et bien évidemment, comme les traducteurs francophones de GNOME sont tout aussi feignants surbookés que moi, les notes de version n’ont pas encore été traduites. La vidéo de présentation n’est pas prête non plus. À mon avis, ils feraient mieux de s’accorder quelques jours de plus pour qu’au moment de la sortie, tout soit prêt.

GNOME 3.30

Mais revenons dans le vif du sujet. Les utilisateurs de vieilles brouettes seront heureux d’apprendre que durant ce cycle, l’accent a été mis sur l’amélioration des performances. Cette nouvelle version consomme donc beaucoup moins de ressources système, tout en étant bien plus fluide et réactive.

Le partage d’écran a également été amélioré. Un nouveau menu système fait son apparition, affichant un indicateur de connexion distante active, facilitant l’arrêt de la session une fois celle-ci terminée.

GNOME devant servir d’environnement par défaut pour le smartphone Librem 5 qui doit sortir début 2019, de gros efforts ont été fait pour que les différentes applications puissent s’adapter à des résolutions bien plus petites (ce qui bénéficiera également aux mini-machines).

Autre changement important, l’abandon progressif du « menu de l’application » (AppMenu) au profit du menu hamburger que tout le monde connaît bien.

Bien que toujours en bêta, ceux qui voudraient retrouver leurs icônes sur le bureau pourront se rabattre sur l’extension Desktop Icons (qui est développée par le développeur principal de Fichiers).

Pour conclure sur l’interface, si vous avez activé la sur-amplification du son, lorsque vous augmenterez le volume, l’OSD affichera désormais visuellement le dépassement des 100%.


Si vous avez installé des Flatpaks (applications exécutée dans un  « bac à sable », ce qui offre une meilleure sécurité), ces derniers seront désormais mis à jour automatiquement (comportement bien évidemment désactivable). Vous pourrez également demander à être notifié de telles mises à jour.

Même si ce n’est pas propre à GNOME, n’ayant pas publié d’article à ce sujet, j’en profite pour rappeler qu’après plusieurs années de développement, la technologie Flatpak est récemment sortie en version 1.0. Et dans la foulée, le « magasin » d’applications Flathub a également quitté son statut de bêta.


Commençons par Fichiers. La barre d’emplacement a été complètement retravaillée pour rendre plus simples certaines actions. Un bouton a été ajouté dans les propriétés du dossier pour pouvoir ouvrir l’utilitaire Disques.

Application Disques qui prend désormais en charge TrueCrypt et VeraCrypt (il en va d’ailleurs de même de GVFS).

Les paramètres de GNOME bénéficient d’un nouveau panneau pour la gestion des périphériques Thunderbolt. Les panneaux relatifs au matériel sont désormais dynamiques et ne s’afficheront qu’en cas de présence du matériel concerné.

L’application Games, qui vous permet de gérer votre ludothèque, prend désormais en charge les jeux Steam sous forme de Flatpaks. Les jeux Virtual Boy sont également pris en charge. Et si vous installez Games sous forme de Flatpak, vous aurez droit à une prise en charge par défaut des consoles Sega Game Gear et Master System, ainsi que Nintendo DS et Virtual Boy. Niveau fonctionnalités, vous pouvez désormais naviguer dans votre collection avec votre manette et parcourir vos jeux par nom, développeur ou plateforme. Les pochettes sont désormais mises en cache sur le disque, ce qui permet un chargement bien plus rapide de votre collection. Une option pour voir les différents raccourcis clavier a également été ajoutée.

L’Agenda peut désormais suivre le mode nuit de GNOME.

L’application Boxes, qui permet de gérer des machines virtuelles, mais également de se connecter à des systèmes distants (fonctionnalité qu’on oublie trop souvent), prend désormais en charge les protocoles VNC et RDP. Autre nouveauté, la possibilité d’importer des fichiers OVF / OVA, ce qui devrait faciliter le partage de machines virtuelles.

L’environnement de développement Builder a droit à tout plein de nouveautés, mais ça sonne chiant. Puis comme les développeurs sont bien évidemment bilingues, je vais les laisser regarder par eux-mêmes le fichier NEWS ğŸ˜�

Ajustements permet désormais de centrer les nouvelles fenêtres et d’afficher ou non les jours de la semaine dans l’horloge de la barre supérieure.

L’éditeur de texte Gedit annonce s’être concentré sur la simplification de l’expérience utilisateur : unification des menus, suppression des redondances, actions plus explicites, ajout de raccourcis pour la plupart des fonctionnalités disponibles et amélioration de la documentation.

Le navigateur Web offre un mode lecture.

On notera également l’apparition d’une nouvelle application, Podcasts, qui, comme son nom le laisse supposer, vous permet de gérer vos podcasts ğŸ˜�

D’ailleurs, lors du précédent cycle, nous avions également eu droit à une nouvelle application, Utilisation. Cette dernière m’avait vendu du rêve. Voir plus simplement l’utilisation des ressources système (oui, je fais de plus en plus ma feignasse). Alors plutôt légère, on nous avait annoncé l’arrivé prochaine de nouvelles fonctionnalités (activité disque, consommation de la batterie…). Mais malheureusement, comme bien d’autres applications, nous n’avons pas droit à la moindre nouveauté 😭

Pour finir, n’oublions pas les sempiternelles corrections de bugs et autres petites améliorations par-ci par-là. La prise en charge de Wayland continue de s’améliorer et sera peut être finalement utilisable un jour (en ce qui me concerne, tant qu’un plantage du Shell ne pourra pas être rattrapé, comme c’est actuellement le cas avec X11, faisant planter par la même occasion toutes les applications et ainsi perdre le travail en cours, un tel changement est bien évidemment impensable).

La version 3.32, qui devrait être la première à utiliser GTK+ 4, sortira, si tout va bien, le 13 mars 2019.

Il n’y a maintenant plus qu’à attendre que ce soit proposé dans les différentes distributions. J’imagine que la première sera Arch Linux, suivie par Fedora et Ubuntu au mois d’octobre.

23 March 2018

The Great Gatsby and onboarding new contributors

I am re-reading “The Great Gatsby” – my high-school son is studying it in English, and I would like to be able to discuss it with him with the book fresh in my mind –  and noticed this passage in the first chapter which really resonated with me.

…I went out to the country alone. I had a dog — at least I had him for a few days until he ran away — and an old Dodge and a Finnish woman, who made my bed and cooked breakfast and muttered Finnish wisdom to herself over the electric stove.

It was lonely for a day or so until one morning some man, more recently arrived than I, stopped me on the road.

“How do you get to West Egg village?” he asked helplessly.

I told him. And as I walked on I was lonely no longer. I was a guide, a pathfinder, an original settler. He had casually conferred on me the freedom of the neighborhood.

In particular, I think this is exactly how people feel the first time they can answer a question in an open source community for the first time. A switch is flipped, a Rubicon is crossed. They are no longer new, and now they are in a space which belongs, at least in part, to them.

04 December 2017

When faster WiFi means unusable connection

I recently moved home and got FTTC with PlusNet, the speed is good when measuring (almost the advertised 80Mb/20Mb) but the connection was unusable due to TCP connections hanging every few minutes (very annoying with ssh but screen helps, worse when using a website for a payment and needing to retry and trust you will only be charged once).

Yesterday I decided to sit down and investigate. Router has logs which were quite helpful. A lot of things like OUT: BLOCK [9] Packet invalid in connection (Invalid tcp flags for current tcp state: TCP []:54426-​>[]:443 on ppp3)

This followed the laptop being seen moving from interface ath10 to interface ath00 and it was moving back and forth quite often.

Looking at the logs on one of the laptops those switches looked like wlan0: disconnect from AP b8:d9:4d:41:76:fb for new auth to b8:d9:4d:41:76:fa

What happened is that default settings on PlusNet router is to have “identical” 2.4GHz and 5GHz networks so the devices believe they are the same network and switch between AP, but they are actually different and the connection tracking gets reset each time such switch happens.

Disabling the 5GHz network made my connection usable, I could probably just change its settings to make it separate.

07 July 2017


In November 2012 I started running an irregular rebuild of all Mageia packages on x86_64, discarding the built packages, to just detect build breakages.

At first it was running a few times a month, now once a week, except before releases where I run it twice a week.

For the first attempt, on 2012-11-28, we had 10949 packages and 1104 failed to build (10%), by the time of the second attempt 3 weeks later we were down to 6.7% of build failures.

Since then, the distribution has been growing and this has helped detecting packages that needs to be fixed early, really helping for the mass rebuild of each release.

  • Mageia 3 was released on 2013-05-01 with 133 failures out of 11008 (1.2%), most of them being random failures due to the use of make -jN.
  • Mageia 4 was released on 2014-02-01 with 24 failures out of 11739 (0.2%)
  • Mageia 5 was released on 2015-06-20 with 42 failures out of 12455 (0.2%)
  • Mageia 6 is going to be released on 2017-07-XX with 0 failures out of 13650 (0%)!

For those interested in the technical details, I am using iurt (the same as we use on Mageia build system) to rebuild everything, creating a new chroot for each package.

I build 16 packages at once, with -j4, on a virtual machine having 32 cores and 200G ram which I use as tmpfs for the builds.

A full rebuild takes about 20 hours.

13 June 2017

Synology PhotoStation password vulnerability

On Synology NAS, synophoto_dsm_user executable, part of PhotoStation package, was leaking NAS user password on the command line.

Using a simple shell loop to run "ps ax | grep synophoto_dsm_user", it was possible to get user and password credentials for user on the NAS who had PhotoStation enabled with their DSM credentials.

Fortunately, by default, shell access on the NAS is not available (by ssh or telnet), it has to be enabled by the admin.

Still, it is a bad practise to pass credentials to process using command line, which can be intercepted.

PhotoStation version 6.7.1-3419 or earlier is vulnerable. I've contacted Synology and they should release a security fix really shortly, as well as a CVE for it.

Update (June 13, 2017): Synology has released a CVE and the vulnerability is fixed in PhotoStation 6.7.2-3429 or later. Remember to update this package on your NAS !

27 February 2017

Hackweek projet: Let's Encrypt DNS-01 validation for acme.sh with Gandi LiveDNS

Last week was SUSE Hackweek and one of my projects was to get Let's Encrypt configured and working on my NAS.

Let's Encrypt is a project aimed at providing SSL certificates for free, in an automated way.

I wanted to get a SSL certificate for my Synology NAS. Synology now supports natively Let's Encrypt but only if the NAS accepts incoming HTTP / HTTPS connections (which is not always what you want).

Fortunately, the protocol used by Let's Encrypt to validate a hostname (and generate a certificate), Automatic Certificate Management Environment (ACME) has a alternative validation path, DNS-01, based on DNS.

DNS-01 requires access to your DNS server, so you can add a validation token used by Let's Encrypt server, to ensure you own the domain name you are requesting a certificate for.

There is a lot of ACME implementations, but very few supports DNS-01 validation with my DNS provider (gandi.net).

I ended-up using acme.sh, fully written in shell script and tried to plug Gandi DNS support in it.

After some tests, I discovered Gandi current DNS service is not allowing fast changing DNS zone informations (which is somehow a requirement for DNS-01 validation). Fortunately, Gandi is now providing a new LiveDNS server, available in beta, with a RESTful HTTP API.

I was able to get it working quite rapidly with curl, and once the prototype was working, I've cleaned everything and created a pull request for integrating the support in acme.sh.

Now, my NAS has its own Let's Encrypt certificate and will update it every 90 days automatically. Getting and installing a certificate for another server (running openSUSE Leap) only took me 5 minutes.

This was a pretty productive hackweek !

01 February 2017


Février arrive, et le FOSDEM aussi ! L'occasion de revoir des amis, boire des bières et manger des carbonnades flamandes. Tiens, justement une recette qui aurait toute sa place dans GNOME Recipes. Cette nouvelle application aura d'ailleurs une petite conférence de présentation (20min). N'hésitez pas non plus à passer au stand GNOME poser vos questions ou faire un coucou, ou prendre des stickers ;)

25 July 2016

Karlsruhe, here we go !

Because I'm missing my GNOME friends, and because GUADEC is the best conference in the world :)

25 May 2016

GStreamer Spring Hackfest 2016

After missing the last few GStreamer hackfests I finally managed to attend this time. It was held in Thessaloniki, Greece’s second largest city. The city is located by the sea side and the entire hackfest and related activities were either directly by the sea or just a couple blocks away.

Collabora was very well represented, with Nicolas, Mathieu, Lubosz also attending.

Nicolas concentrated his efforts on making kmssink and v4l2dec work together to provide zero-copy decoding and display on a Exynos 4 board without a compositor or other form of display manager. Expect a blog post soon  explaining how to make this all fit together.

Lubosz showed off his VR kit. He implemented a viewer for planar point clouds acquired from a Kinect. He’s working on a set of GStreamer plugins to play back spherical videos. He’s also promised to blog about all this soon!

Mathieu started the hackfest by investigating the intricacies of Albanian customs, then arrived on the second day in Thessaloniki and hacked on hotdoc, his new fancy documentation generation tool. He’ll also be posting a blog about it, however in the meantime you can read more about it here.

As for myself, I took the opportunity to fix a couple GStreamer bugs that really annoyed me. First, I looked into bug #766422: why glvideomixer and compositor didn’t work with RTSP sources. Then I tried to add a ->set_caps() virtual function to GstAggregator, but it turns out I first needed to delay all serialized events to the output thread to get predictable outcomes and that was trickier than expected. Finally, I got distracted by a bee and decided to start porting the contents of docs.gstreamer.com to Markdown and updating it to the GStreamer 1.0 API so we can finally retire the old GStreamer.com website.

I’d also like to thank Sebastian and Vivia for organising the hackfest and for making us all feel welcomed!

GStreamer Hackfest Venue

25 May 2015

SUSE Ruling the Stack in Vancouver

Rule the Stack

Last week during the the OpenStack Summit in Vancouver, Intel organized a Rule the Stack contest. That's the third one, after Atlanta a year ago and Paris six months ago. In case you missed earlier episodes, SUSE won the two previous contests with Dirk being pretty fast in Atlanta and Adam completing the HA challenge so we could keep the crown. So of course, we had to try again!

For this contest, the rules came with a list of penalties and bonuses which made it easier for people to participate. And indeed, there were quite a number of participants with the schedule for booking slots being nearly full. While deploying Kilo was a goal, you could go with older releases getting a 10 minutes penalty per release (so +10 minutes for Juno, +20 minutes for Icehouse, and so on). In a similar way, the organizers wanted to see some upgrade and encouraged that with a bonus that could significantly impact the results (-40 minutes) — nobody tried that, though.

And guess what? SUSE kept the crown again. But we also went ahead with a new challenge: outperforming everyone else not just once, but twice, with two totally different methods.

For the super-fast approach, Dirk built again an appliance that has everything pre-installed and that configures the software on boot. This is actually not too difficult thanks to the amazing Kiwi tool and all the knowledge we have accumulated through the years at SUSE about building appliances, and also the small scripts we use for the CI of our OpenStack packages. Still, it required some work to adapt the setup to the contest and also to make sure that our Kilo packages (that were brand new and without much testing) were fully working. The clock result was 9 minutes and 6 seconds, resulting in a negative time of minus 10 minutes and 54 seconds (yes, the text in the picture is wrong) after the bonuses. Pretty impressive.

But we also wanted to show that our product would fare well, so Adam and I started looking at this. We knew it couldn't be faster than the way Dirk picked, and from the start, we targetted the second position. For this approach, there was not much to do since this was similar to what he did in Paris, and there was work to update our SUSE OpenStack Cloud Admin appliance recently. Our first attempt failed miserably due to a nasty bug (which was actually caused by some unicode character in the ID of the USB stick we were using to install the OS... we fixed that bug later in the night). The second attempt went smoother and was actually much faster than we had anticipated: SUSE OpenStack Cloud deployed everything in 23 minutes and 17 seconds, which resulted in a final time of 10 minutes and 17 seconds after bonuses/penalties. And this was with a 10 minutes penalty due to the use of Juno (as well as a couple of minutes lost debugging some setup issue that was just mispreparation on our side). A key contributor to this result is our use of Crowbar, which we've kept improving over time, and that really makes it easy and fast to deploy OpenStack.

Wall-clock time for SUSE OpenStack Cloud

Wall-clock time for SUSE OpenStack Cloud

These two results wouldn't have been possible without the help of Tom and Ralf, but also without the whole SUSE OpenStack Cloud team that works on a daily basis on our product to improve it and to adapt it to the needs of our customers. We really have an awesome team (and btw, we're hiring)!

For reference, three other contestants succeeded in deploying OpenStack, with the fastest of them ending at 58 minutes after bonuses/penalties. And as I mentioned earlier, there were even more contestants (including some who are not vendors of an OpenStack distribution), which is really good to see. I hope we'll see even more in Tokyo!

Results of the Rule the Stack contest

Results of the Rule the Stack contest

Also thanks to Intel for organizing this; I'm sure every contestant had fun and there was quite a good mood in the area reserved for the contest.

Update: See also the summary of the contest from the organizers.

12 May 2015

Deploying Docker for OpenStack with Crowbar

A couple of months ago, I was meeting colleagues of mine working on Docker and discussing about how much effort it would be to add support for it to SUSE OpenStack Cloud. It's been something that had been requested for a long time by quite a number of people and we never really had time to look into it. To find out how difficult it would be, I started looking at it on the evening; the README confirmed it shouldn't be too hard. But of course, we use Crowbar as our deployment framework, and the manual way of setting it up is not really something we'd want to recommend. Now would it be "not too hard" or just "easy"? There was only way to know that... And guess what happened next?

It took a couple of hours (and two patches) to get this working, including the time for packaging the missing dependencies and for testing. That's one of the nice things we benefit from using Crowbar: adding new features like this is relatively straight-forward, and so we can enable people to deploy a full cloud with all of these nice small features, without requiring them to learn about all the technologies and how to deploy them. Of course this was just a first pass (using the Juno code, btw).

Fast-forward a bit, and we decided to integrate this work. Since it was not a simple proof of concept anymore, we went ahead with some more serious testing. This resulted in us backporting patches for the Juno branch, but also making Nova behave a bit better since it wasn't aware of Docker as an hypervisor. This last point is a major problem if people want to use Docker as well as KVM, Xen, VMware or Hyper-V — the multi-hypervisor support is something that really matters to us, and this issue was actually the first one that got reported to us ;-) To validate all our work, we of course asked tempest to help us and the results are pretty good (we still have some failures, but they're related to missing features like volume support).

All in all, the integration went really smoothly :-)

Oh, I forgot to mention: there's also a docker plugin for heat. It's now available with our heat packages now in the Build Service as openstack-heat-plugin-heat_docker (Kilo, Juno); I haven't played with it yet, but this post should be a good start for anyone who's curious about this plugin.

15 April 2015

Be IP is hiring!

In case some readers of this blog would be interested in working with Open Source software and VoIP technologies, Be IP (http://www.beip.be) is hiring a developer. Please see http://www.beip.be/BeIP-Job-Offer.pdf for the job description. You can contact me directly.

25 January 2015

Ekiga 5 – Progress Report

Current Status Ekiga 5 has progressed a lot lately. OpenHUB is reportin a High Activity for the project. The main reason behind this is that I am again dedicating much of my spare time to the project. Unfortunately, we are again facing a lack of contributions. Most probably (among others) because the project has been […]

15 August 2014

GNOME.Asia Summit 2014

Everyone has been blogging about GUADEC, but I’d like to talk about my other favorite conference of the year, which is GNOME.Asia. This year, it was in Beijing, a mightily interesting place. Giant megapolis, with grandiose architecture, but at the same time, surprisingly easy to navigate with its efficient metro system and affordable taxis. But the air quality is as bad as they say, at least during the incredibly hot summer days where we visited.

The conference itself was great, this year, co-hosted with FUDCon’s asian edition, it was interesting to see a crowd that’s really different from those who attend GUADEC. Many more people involved in evangelising, deploying and using GNOME as opposed to just developing it, so it allows me to get a different perspective.

On a related note, I was happy to see a healthy delegation from Asia at GUADEC this year!

Sponsored by the GNOME Foundation

25 March 2013

SPICE on OSX, take 2

A while back, I made a Vinagre build for OSX. However, reproducing this build needed lots of manual tweaking, the build was not working on newer OSX versions, and in the mean time, the recommended SPICE client became remote-viewer. In short, this work was obsolete.

I've recently looked again at this, but this time with the goal of documenting the build process, and making the build as easy as possible to reproduce. This is once again based off gtk-osx, with an additional moduleset containing the SPICE modules, and a script to download/install most of what is needed. I've also switched to building remote-viewer instead of vinagre

This time, I've documented all of this work, but all you should have to do to build remote-viewer for OSX is to run a script, copy a configuration file to the right place, and then run a usual jhbuild build. Read the documentation for more detailed information about how to do an OSX build.

I've uploaded a binary built using these instructions, but it's lacking some features (USB redirection comes to mind), and it's slow, etc, etc, so .... patches welcome! ;) Feel free to contact me if you are interested in making OSX builds and need help getting started, have build issues, ...

11 December 2012

FOSDEM 2013 Crossdesktop devroom Call for talks

The Call for talks for the Crossdesktop devroom at FOSDEM 2013 is getting to its end this Friday. Don't wait and submit your talk proposal about your favourite part of GNOME now!

Proposals should be sent to the crossdesktop devroom mailing list (you don't have to subscribe).

04 July 2011

Going to RMLL (LSM) and Debconf!

Next week, I’ll head to Strasbourg for Rencontres Mondiales du Logiciel Libre 2011. On monday morning, I’ll be giving my Debian Packaging Tutorial for the second time. Let’s hope it goes well and I can recruit some future DDs!

Then, at the end of July, I’ll attend Debconf again. Unfortunately, I won’t be able to participate in Debcamp this year, but I look forward to a full week of talks and exciting discussions. There, I’ll be chairing two sessions about Ruby in Debian and Quality Assurance.

17 February 2011

Recent Libgda evolutions

It’s been a long time since I blogged about Libgda (and for the matter since I blogged at all!). Here is a quick outline on what has been going on regarding Libgda for the past few months:

  • Libgda’s latest version is now 4.2.4
  • many bugs have been corrected and it’s now very stable
  • the documentation is now faily exhaustive and includes a lot of examples
  • a GTK3 branch is maintained, it contains all the modifications to make Libgda work in the GTK3 environment
  • the GdaBrowser and GdaSql tools have had a lot of work and are now both mature and stable
  • using the NSIS tool, I’ve made available a new Windows installer for the GdaBrowser and associated tools, available at http://www.gnome.org/~vivien/GdaBrowserSetup.exe. It’s only available in English and French, please test it and report any error.

In the next months, I’ll work on polishing even more the GdaBrowser tool which I use on a daily basis (and of course correct bugs).

16 March 2010

Webkit fun, maths and an ebook reader

I have been toying with webkit lately, and even managed to do some pretty things with it. As a consequence, I haven’t worked that much on ekiga, but perhaps some of my experiments will turn into something interesting there. I have an experimental branch with a less than fifty lines patch… I’m still trying to find a way to do more with less code : I want to do as little GObject-inheritance as possible!

That little programming was done while studying class field theory, which is pretty nice on the high-level principles and somewhat awful on the more technical aspects. I also read again some old articles on modular forms, but I can’t say that was “studying” : since it was one of the main objects of my Ph.D, that came back pretty smoothly…

I found a few minutes to enter a brick-and-mortar shop and have a look at the ebook readers on display. There was only *one* of them : the sony PRS-600. I was pretty unimpressed : the display was too dark (because it was a touch screen?), but that wasn’t the worse deal breaker. I inserted an SD card where I had put a sample of the type of documents I read : they showed up as a flat list (pain #1), and not all of them (no djvu) (pain #2) and finally, one of them showed up too small… and ended up fully unreadable when I tried to zoom (pain #3). I guess that settles the question I had on whether my next techno-tool would be a netbook or an ebook reader… That probably means I’ll look more seriously into fixing the last bug I reported on evince (internal bookmarks in documents).

16 January 2010

New Libgda releases

With the beginning of the year comes new releases of Libgda:

  • version 4.0.6 which contains corrections for the stable branch
  • version 4.1.4, a beta version for the upcoming 4.2 version

The 4.1.4’s API is now considered stable and except for minor corrections should not be modified anymore.

This new version also includes a new database adaptator (provider) to connect to databases through a web server (which of course needs to be configured for that purpose) as illustrated by the followin diagram:

WebProvider usage

The database being accessed by the web server can be any type supported by the PEAR::MDB2 module.

The GdaBrowser application now supports defining presentation preferences for each table’s column, which are used when data from a table’s column need to be displayed:
GdaBrowser table column's preferences
The UI extension now supports improved custom layout, described through a simple XML syntax, as shown in the following screenshot of the gdaui-demo-4.0 program:

Form custom layout

For more information, please visit the http://www.gnome-db.org web site.

05 November 2009

Attracted to FLT

I have been a little stuck for some weeks : a new year started (no, that post hasn’t been stuck since january — scholar year start in september) and I have students to tend to. As I have the habit to say : good students bring work because you have to push them high, and bad students bring work because you have to push them from low! Either way, it has been keeping me pretty busy.

Still, I found the time to read some more maths, but got lost on something quite unrelated to my main objective : I just read about number theory and the ideas behind the proof of Fermat’s Last Theorem (Taylor and Wiles’ theorem now). That was supposed to be my second target! Oh, well, I’ll just try to hit my first target now (Deligne’s proof of the Weil conjectures). And then go back to FLT for a new and deeper reading.

I only played a little with ekiga’s code — mostly removing dead code. Not much : low motivation.

11 July 2009

Slides from RMLL (and much more)

So, I’m back from the Rencontres Mondiales du Logiciel Libre, which took place in Nantes this year. It was great to see all those people from the french Free Software community again, and I look forward to seeing them again next year in Bordeaux (too bad the Toulouse bid wasn’t chosen).

The Debian booth, mainly organized by Xavier Oswald and Aurélien Couderc, with help from Raphaël, Roland and others (but not me!), got a lot of visits, and Debian’s popularity is high in the community (probably because RMLL is mostly for über-geeks, and Debian’s market share is still very high in this sub-community).

I spent quite a lot of time with the Ubuntu-FR crew, which I hadn’t met before. They do an awesome work on getting new people to use Linux (providing great docs and support), and do very well (much better than in the past) at giving a good global picture of the Free Software world (Linux != Ubuntu, other projects do exist and play a very large role in Ubuntu’s success, etc). It’s great to see Free Software’s promotion in France being in such good hands. (Full disclosure: I got a free mug (recycled plastic) with my Ubuntu-FR T-shirt, which might affect my judgement).

I gave two talks, on two topics I wanted to talk about for some time. First one was about the interactions between users, distributions and upstream projects, with a focus on Ubuntu’s development model and relationships with Debian and upstream projects. Second one was about voting methods, and Condorcet in particular. If you attended one of those talks, feedback (good or bad) is welcomed (either in comments or by mail). Slides are also available (in french):

On a more general note, I still don’t understand why the “Mondiales” in RMLL’s title isn’t being dropped or replaced by “Francophones“. Seeing the organization congratulate themselves because 30% of the talks were in english was quite funny, since in most cases, the english part of the talk was “Is there someone not understanding french? no? OK, let’s go on in french.“, and all the announcements were made in french only. Seriously, RMLL is a great (probably the best) french-speaking community event. But it’s not FOSDEM: different goals, different people. Instead of trying (and failing) to make it an international event, it would be much better to focus on making it a better french-speaking event, for example by getting more french-speaking developers to come and talk (you see at least 5 times more french-speaking developers in FOSDEM than in RMLL).

I’m now back in Lyon for two days, before leaving to Montreal Linux Symposium, then coming back to Lyon for three days, then Debconf from 23rd to 31st, and then moving to Nancy, where I will start as an assistant professor in september (a permanent (tenured) position).

22 July 2008

Looking for a job

On September I finish my studies of computer science, so I start to search a job. I really enjoyed my current job at Collabora maintaining Empathy, I learned lots of things about the Free Software world and I would like to keep working on free software related projects if possible. My CV is available online here.

Do you guys know any company around the free software and GNOME looking for new employees? You can contact me by email to xclaesse@gmail.com

22 April 2008

Enterprise Social Search slideshow

Enterprise Social Search is a way to search, manage, and share information within a company. Who can help you find relevant information and nothing but relevant information? Your colleagues, of course

Today we are launching at Whatever (the company I work for) a marketing campaign for our upcoming product: Knowledge Plaza. Exciting times ahead!

03 November 2007

git commit / darcs record

I’ve been working wit git lately but I have also missed the darcs user interface. I honestly think the darcs user interface is the best I’ve ever seen, it’s such a joy to record/push/pull (when darcs doesn’t eat your cpu) 🙂

I looked at git add --interactive because it had hunk-based commit, a pre-requisite for darcs record-style commit, but it has a terrible user interface, so i just copied the concept: running a git diff, filtering hunks, and then outputing the filtered diff through git apply --cached.

It supports binary diffs, file additions and removal. It also asks for new files to be added even if this is not exactly how darcs behave but I always forget to add new files, so I added it. It will probably break on some extreme corner cases I haven’t been confronted to, but I gladly accept any patches 🙂

Here’s a sample session of git-darcs-record script:

$ git-darcs-record
Add file:  newfile.txt
Shall I add this file? (1/1) [Ynda] : y

Binary file changed: document.pdf

Shall I record this change? (1/7) [Ynda] : y

@@ -1,3 +1,5 @@

Shall I record this change? (2/7) [Ynda] : y

@@ -1,17 +1,5 @@
 #!/usr/bin/env python

-# git-darcs-record, emulate "darcs record" interface on top of a git repository
-# Usage:
-# git-darcs-record first asks for any new file (previously
-#    untracked) to be added to the index.
-# git-darcs-record then asks for each hunk to be recorded in
-#    the next commit. File deletion and binary blobs are supported
-# git-darcs-record finally asks for a small commit message and
-#    executes the 'git commit' command with the newly created
-#    changeset in the index
 # Copyright (C) 2007 Raphaël Slinckx
 # This program is free software; you can redistribute it and/or

Shall I record this change? (3/7) [Ynda] : y

@@ -28,6 +16,19 @@
 # along with this program; if not, write to the Free Software
 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.

+# git-darcs-record, emulate "darcs record" interface on top of a git repository
+# Usage:
+# git-darcs-record first asks for any new file (previously
+#    untracked) to be added to the index.
+# git-darcs-record then asks for each hunk to be recorded in
+#    the next commit. File deletion and binary blobs are supported
+# git-darcs-record finally asks for a small commit message and
+#    executes the 'git commit' command with the newly created
+#    changeset in the index
 import re, pprint, sys, os

 BINARY = re.compile("GIT binary patch")

Shall I record this change? (4/7) [Ynda] : n

@@ -151,16 +152,6 @@ def read_answer(question, allowed_responses=["Y", "n", "d", "a"]):
        return resp

-def setup_git_dir():
-       global GIT_DIR
-       GIT_DIR = os.getcwd()
-       while not os.path.exists(os.path.join(GIT_DIR, ".git")):
-               GIT_DIR = os.path.dirname(GIT_DIR)
-               if GIT_DIR == "/":
-                       return False
-       os.chdir(GIT_DIR)
-       return True
 def git_get_untracked_files():

Shall I record this change? (5/7) [Ynda] : y

# On branch master
# Changes to be committed:
#   (use "git reset HEAD file..." to unstage)
#       modified:   document.pdf
#       modified:   foobar.txt
#       modified:   git-darcs-record
#       new file:   newfile.txt
# Changed but not updated:
#   (use "git add file file..." to update what will be committed)
#       modified:   git-darcs-record
What is the patch name? Some cute patch name
Created commit a08f34e: Some cute patch name
 4 files changed, 3 insertions(+), 29 deletions(-)
 create mode 100644 newfile.txt

Get the script here: git-darcs-record script and put in somewhere in your $PATH. Any comments or improvements is welcome !

22 January 2007

Un nouveau laptop, sans windows !

Voilà, j’y pensais depuis longtemps et c’est maintenant chose faite, je me suis acheté un tout nouveau ordinateur portable.

Je l’ai acheté sur le site français LDLC.com et me suis renseigné pour savoir si il était possible d’acheter les ordinateurs de leur catalogue sans logiciels (principalement sans windows). Je leur ai donc envoyé un email, et à ma grande surprise ils m’on répondu que c’était tout a fait possible, qu’il suffi de passer commande et d’envoyer ensuite un email pour demander de supprimer les logiciels de la commande. J’ai donc commandé mon laptop et ils m’ont remboursé de 20€ pour les logiciels, ce n’est pas énorme sur le prix d’un portable, mais symboliquement c’est déjà ça.

Toutes fois je me pose des questions, pourquoi cette offre n’est pas inscrite sur le site de LDLC ? En regardant sous mon tout nouveau portable je remarque une chose étrange, les restes d’un autocollant qu’on a enlevé, exactement à l’endroit où habituellement est collé la clef d’activation de winXP. Le remboursement de 20€ tout rond par LDLC me semble également étrange vue que LDLC n’est qu’un intermédiaire, pas un constructeur, et donc eux achètent les ordinateurs avec windows déjà installé. Bref tout ceci me pousse à croire que c’est LDLC qui perd les 20€ et je me demande dans quel but ?!? Pour faire plaisir aux clients libre-istes ? Pour éviter les procès pour vente liée ? Pour à leur tours se faire rembourser les licences que les clients n’ont pas voulu auprès du constructeur/Microsoft et éventuellement gagner plus que 20€ si les licences OEM valent plus que ça ? Bref ceci restera sans doutes toujours un mistère.

J’ai donc installé Ubuntu qui tourne plutôt bien. J’ai été même très impressionné par le network-manager qui me connecte automatiquement sur les réseaux wifi ou filaire selon la disponibilité et qui configure même un réseau zeroconf si il ne trouve pas de server dhcp, c’est très pratique pour transférer des données entre 2 ordinateurs, il suffi de brancher un cable ethernet (ça marche aussi par wifi mais j’ai pas encore testé) entre les 2 et hop tout le réseau est configuré automatiquement sans rien toucher, vraiment magique ! Windows peut aller se cacher, ubuntu est largement plus facile d’utilisation !

20 December 2006

Documenting bugs

I hate having to write about bugs in the documentation. It feels like waving a big flag that says ‘Ok, we suck a bit’.

Today, it’s the way fonts are installed, or rather, they aren’t. The Fonts folder doesn’t show the new font, and the applications that are already running don’t see them.

So I’ve fixed the bug that was filed against the documentation. Now it’s up to someone else to fix the bugs in Gnome.

05 December 2006

Choice and flexibility: bad for docs

Eye of Gnome comes with some nifty features like support for EXIF data in jpegs. But this depends on a library that isn’t a part of Gnome.

So what do I write in the user manual for EOG?

‘You can see EXIF data for an image, but you need to check the innards of your system first.’
‘You can maybe see EXIF data. I don’t know. Ask your distro.’
‘If you can’t see EXIF data, install the libexif library. I’m sorry, I can’t tell you how you can do that as I don’t know what sort of system you’re running Gnome on.’

The way GNU/Linux systems are put together is perhaps great for people who want unlimited ability to customize and choose. But it makes it very hard to write good documentation. In this sort of scenario, I would say it makes it impossible, and we’re left with a user manual that looks bad.

I’ve added this to the list of use cases for Project Mallard, but I don’t think it’ll be an easy one to solve.


Planète GNOME-FR

Planète GNOME-FR est un aperçu de la vie, du travail et plus généralement du monde des membres de la communauté GNOME-FR.

Certains billets sont rédigés en anglais car nous collaborons avec des gens du monde entier.

Dernière mise à jour :
17 February 2020 à 12:58 UTC
Toutes les heures sont UTC.


Planète GNOME-FR est propulsée par l'agrégateur Planet, cron, Python, Red Hat (qui héberge ce serveur).

Le design du site est basé sur celui des sites GNOME et de Planet GNOME.

Planète GNOME-FR est maintenue par Frédéric Péters et Luis Menina. Si vous souhaitez ajouter votre blog à cette planète, il vous suffit d'ouvrir un bug. N'hésitez pas à nous contacter par courriel pour toute autre question.