You are here: Home / Blog / IRILL blog

IRILL blog

Blog of IRILL (Research and Innovation Initiative on Free Software)

Polyhedral compilation for the masses

The PARKAS (INRIA and ENS/DI), PEQUAN (UPMC/LIP6) teams and IRILL are glad to invite you to a mini-workshop bringing together researchers in parallelizing compilation, polyhedral algorithms, scientific computing on GPUs, and formal methods for signal-processing and control systems.


Universite Pierre et Marie Curie
Room 25-26/105 (first floor, enter through tower 26)

Tuesday October 21st
Mini-workshop:   9:30 - 12:15
Thesis defense: 14:15 - ?

For logistical purposes, please send an email before October 14
if you plan to attend the mini-workshop:

Preliminary program

09:30-09:35   Welcome

09:35-10:10 - P. Sadayappan, Ohio State U.
10:10-10:45 - Sven Verdoolaege, INRIA and K. U. Leuven

10:45-11:00   Break

11:00-11:35 - Franz Franchetti, Carnegie Mellon U.
11:35-12:10 - Issam Said, U. Pierre et Marie Curie

14:15-      - Thesis defense, Tobias Grosser, INRIA

Titles and abstracts

Speaker: P. Sadayappan, Ohio State U.
Title: On Other Uses for the Polyhedral Model Besides Performance

The use of the polyhedral model is now widely acknowledged as the
most effective approach to compiler optimization of affine loop
computations. However, a significant unsolved challenge is that of
effectively using the polyhedral model to optimize real application
codes, which rarely are fully affine, but invariably also contain
interleaved non-affine computations. In this talk, two uses of the
polyhedral model for other goals than loop transformations will be
discussed: 1) a compiler-assisted approach to detecting transient
memory errors, and 2) compiler-assistance for software cache
coherence. With both these uses, arbitrarily interleaved
affine/non-affine codes can be handled, where the polyhedral
analysis offers benefits for the affine code portions, along with a
conservative handling of the non-affine computations.

Speaker: Sven Verdoolaege, INRIA and K. U. Leuven
Title: Integer Set Coalescing for Polyhedral Compilation

Polyhedral compilation is widely used in high-level synthesis tools
and in production compilers such as gcc, LLVM and IBM/XL.  It is
based on the polyhedral model, a powerful abstraction for analyzing
and transforming (parts of) programs that are ``sufficiently
regular''.  The key feature of this model is that it is instance
based, allowing for a representation and treatment of individual
dynamic executions of a statement inside a loop nest and/or
individual array elements.  These individual elements are
represented by tuples of integers which are collected in sets and
relations described by affine constraints.  Polyhedral compilation
then mainly consists of an analysis and transformation of these sets
and relations.

There are typically many ways to describe the same set of integer
tuples using affine constraints.  Most operations that can be
performed on these sets do not depend on which particular
representation is used, in the sense that only the time required for
the computation is affected by this choice but not the value of the
result.  There are however some operations where this choice can
have a significant impact, notably the computation of
(overapproximations of) transitive closures and the construction of
an AST for visiting each point in a set.  Assuming a representation
in disjunctive normal form, it is in all cases important to keep the
number of disjuncts as small as possible.  Excess disjuncts may
result from operations such as subtraction, lexicographic
optimization and integer projection.  Set coalescing tries to
combine several disjuncts into a single disjunct without affecting
the elements in the set.  We describe several cases where coalescing
can be applied.

Speaker: Franz Franchetti, Carnegie Mellon U.
Title: High Assurance Spiral: Co-Synthesizing Proof and Implementation
From High-Level Specification

In this talk we introduce "High Assurance SPIRAL" to solve the "last
mile" problem for the synthesis of high assurance implementations of
controllers for vehicular systems that are executed in todays and
future embedded and high performance embedded system
processors. High Assurance SPIRAL is a methodology to translate a
high level specification of a high assurance controller into a
highly resource-efficient, platform-adapted, verified control
software implementation for a given platform in a language like C or
C++. High Assurance SPIRAL proves that the implementation is
equivalent to the specification written in the control engineer's
domain language. Our approach supports problems involving
floating-point calculations and provides highly optimized
synthesized code.  At the core of High Assurance SPIRAL is the
Hybrid Control Operator Language (HCOL) that leverages advanced
mathematical constructs expressing the controller specification to
provide high quality translation capabilities.

Speaker: Issam Said, U. Pierre et Marie Curie
Title: Evaluating APU for high performance scientific computing

AMD APUs eliminate the PCI Express bus which bottlenecks many GPU
scientific applications. However, integrated GPUs in APUs are less
powerful than mainstream discrete GPUs, and while the upcoming APUs
will rely on a unified memory, the first APUs still have a distinct
GPU memory partition. Hence, it is worthwhile to investigate the
applications, as well as problem sizes, for which the GPU part of an
APU may outperform a discrete GPU. In this talk we assess the
relevance of the early generations of APUs for scientific HPC via
hardware and applicative OpenCL micro-benchmarks.  We present
detailed measurements of CPU-GPU data transfers and performance
tests of highly optimized 3D finite difference stencils.  We also
emphasize the importance of data placement when APUs are used.  Our
results show that integrated GPUs of upcoming APUs can outperform
discrete high-end GPUs for medium sized problems with high
communication requirements.

PhD candidate: Tobias Grosser, INRIA
Title: A Decoupled Approach to High-Level Loop Optimization,
tile shapes, polyhedral building blocks and low-level compilers

Despite decades of research on high-level loop optimizations and
their successful integration in production C/C++/FORTRAN compilers,
most compiler internal loop transformation systems only partially
address the challenges posed by the increased complexity and
diversity of today's hardware. Especially when exploiting domain
specific knowledge to obtain optimal code for complex targets such
as accelerators or many-cores processors, many existing loop
optimization frameworks have difficulties exploiting this
hardware. As a result, new domain specific optimization schemes are
developed independently without taking advantage of existing loop
optimization technology. This results both in missed optimization
opportunities as well as low portability of these optimization
schemes to different compilers. One area where we see the need for
better optimizations are iterative stencil computations, an
important computational problem that is regularly optimized by
specialized, domain specific compilers, but where generating
efficient code is difficult.

In this work we present new domain specific optimization strategies
that enable the generation of high-performance GPU code for stencil
computations. Different to how most existing domain specific
compilers are implemented, we decouple the high-level optimization
strategy from the low-level optimization and specialization
necessary to yield optimal performance. As high-level optimization
scheme we present a new formulation of split tiling, a tiling
technique that ensures reuse along the time dimension as well as
balanced coarse grained parallelism without the need for redundant
computations.  Using split tiling we show how to integrate a domain
specific optimization into a general purpose C-to-CUDA translator,
an approach that allows us to reuse existing non-domain specific
optimizations.  We then evolve split tiling into a hybrid
hexagonal/parallelogram tiling scheme that allows to generate code
that even better addresses GPU specific concerns. To conclude our
work on tiling we investigate the relation between diamond and
hexagonal tiling.  Starting with a detailed analysis of diamond
tiling including the requirements it poses on tile sizes and
wavefront coefficients, we provide a unified formulation of
hexagonal and diamond tiling to perform hexagonal tiling for two
dimensional problems (one time, one space) in the context of a
general purpose optimizer such as Pluto. Finally, we use this
formulation to evaluate hexagonal and diamond tiling in terms of
compute-to-communication and compute-to-synchronization ratios.

In the second part of this work, we discuss our contributions to
important infrastructure components, our building blocks, that
enviable us to decouple our high-level optimizations from both the
nec- essary code generation optimizations as well as the compiler
infrastructure we apply the optimization to. We start with
presenting a new polyhedral extractor that obtains a polyhedral
representation from a piece of C code, widening the supported C code
to exploit the full generality of Presburger arithmetic and taking
special care of modeling language semantics even in the presence of
defined integer wrapping. As a next step, we present a new
polyhedral AST generation approach, which extends AST generation
beyond classical control flow generation by allowing the generation
of user provided mappings. Providing a fine-grained option
mechanism, we give the user fine grained control about AST generator
decisions and add extensive support for specialization e.g., with a
new generalized form of polyhedral unrolling. To facilitate the
implementation of polyhedral transformations, we present a new
schedule representation, schedule trees, which proposes to make the
inherent tree structure of schedules explicit to simplify the work
with complex polyhedral schedules.

The last part of this work takes a look at our contributions to
low-level compilers. The main focus in this part is our work on
optimistic delinearization, an approach to derive a
multi-dimensional array view for multi-variate polynomial
expressions which commonly result from code that models data as
multi-dimensional arrays of parametric size.


A software environment whose goal is the formal specification and verification of dynamic systems.


The CosyVerif platform offers a set of services that can be operated on models. Each model must respect the definition of a formalism. Then, when you connect to using a client, you may browse and activate the services that are available for this formalism.

All the developed software are open source and free software tools.

Alligator is published under the GNU Affero General Public License (AGPL) version 3 .

Coloane is published under the Eclipse Public License (EPL) version 1 .


Livrets bleus du logiciel libre

IRILL, par l'intermédiaire du groupe thématique Logiciel Libre, a contribué aux nouveaux livrets bleus.

Les trois livrets publiés traitent des sujets clés portés par IRILL. Ils sont disponibles aux adresses suivantes (PDF):

Plus d'information

Tails HackFest 2014

IRILL is happy to Host the Tails HackFest, on July 5-6, 2014.
Below you can find the full announcement by the Tails community.

Join us at the Tails HackFest, 2014!
July 5-6, 2014 — Paris, France

Description and goals

Join us to make online anonymity and digital privacy usable by the masses! Whether you're a writer, a software developer, a designer, a system administrator or just plain interested, come learn about the challenges faced by Tails, and how you can be part of the solution. The Tails HackFest will bring together anyone interested in making Tails more usable and more secure. This open event will be an intense mix of teaching, drawing, coding, sharing, learning and celebrating.


  • Where: the venue for the event is IRILL, Paris, France (
  • Dates: Saturday, July 5, 2014 - Sunday, July 6, 2014
  • Time: 10 AM - 10 PM
  • Registration: if you want to attend, please consider dropping us a note about it. This is optional, but would help organizing this event.
  • Contact: <>, #tails-hackfest on
  • Details, scheduling and updates:

What is Tails?

Tails is a live operating system that can be started on almost any computer from a DVD, USB stick, or SD card. It is Free Software, and based on Debian GNU/Linux.

Tails provides a platform to solve many surveillance problems by "doing the right thing" out of the box by default, protecting eve less tech-savvy users from the most likely and highest impact risks.

It aims at preserving privacy and anonymity, and helps to:

  • use the Internet anonymously and circumvent censorship; all connections to the Internet are forced to go through the Tor network;
  • leave no trace on the computer being used unless the user asks it explicitly;
  • use state-of-the-art cryptographic tools to encrypt files, emails and instant messaging.

Tails is about usability: every feature and software is ready-to-use, thoroughly documented, and translated into many languages.

Tails is about cooperation: all products are released as Free and Open Source Software, and shared with other projects whenever possible.

People use Tails to write books and create movies. People use Tails to chat off-the-record, browse the web anonymously and share sensitive documents. Many people depend on Tails to do their daily work, if not simply to stay alive.

Looking forward to meet you on July 5-6! No doubt you'll find a great way to contribute to Tails, regardless of what your field of expertise is!

Host and sponsors

Many thanks to Debian, IRILL, Mozilla and the Tor project for supporting this event!

Join us at the Tails HackFest, 2014!Join us to make online anonymity and digital privacy usable by the masses! Whether you're a writer, a software developer, a designer, a system administrator or just plain interested, come learn about the challenges faced by Tails, and how you can be part of the solution. The Tails HackFest will bring together anyone interested in making Tails more usable and more secure. This open event will be an intense mix of teaching, drawing, coding, sharing, learning and celebrating.

Opam Weather Service now online!

Irill starts a package weather service for Opam, and the OCaml community.

Irill is now providing a valuable service to the OCaml community, via a static analysis of the packages present in the Opam repository, based on technology developed in the Mancoosi project.

On it is now possible to find an overview of the OCaml packages that can or cannot be properly installed, with detailed explanations of the reasons for the failures.

More detailed information can be found in this blog article.

Meetup Salt

For the third edition, IRILL is glad to host the Salt meetup. The event will take place Monday 19th, from 19:00 to 23:00.

SaltStack is similar to Chef or Puppet. It is an open source configuration management and remote execution application. More information:

Thomas S Hatch, the original author of Salt, will be present to this event.

More information on the Logilab blog.

This meetup is organized by our friends from Logilab.

SaltStack Paris Meetup

Logilab has set up the second meetup for salt users in Paris on Feb 6th, 2014 at IRILL, near Place d'Italie, starting at 18:00.

Here is the announce in french

Please forward it to whom may be interested, underlining that pizzas will be offered to refuel the chatters ;)

Conveniently placed a week after the Salt Conference, topics will include anything related to salt and its uses, demos, new ideas, exchange of salt formulas, commenting the talks/videos of the saltconf, etc.

If you are interested in Salt, Python and Devops and will be in Paris at that time, we hope to see you there !

Analyse de l'utilisation mémoire des applications OCaml sans changer leur comportement

Thursday January 30th, Cagdas Bozman (OCamlPro - ENSTA) will come at IRILL to talk about "Analyse de l'utilisation mémoire des applications OCaml sans changer leur comportement"

Nous étudions le comportement mémoire des applications OCaml, dans le but de mieux comprendre l'utilisation mémoire et essayer de détecter des problèmes comme les fuites mémoire. Dans l'exposé, nous présenterons un premier outil qui génère des graphes de la mémoire des applications OCaml.

Debian Ruby Team 2014 meeting

IRILL hosts the Debian team in charge of Ruby.

On January 15-17, IRILL will host a meeting of the Debian team in charge of Ruby. Organized as a satellite event of the Paris MiniDebConf, this meeting will be focused on two main goals: the transition toward the removal of ruby1.8 from the Debian archive and quality assurance work around Ruby on Rails Debian packages.

QEMU/CPC: Static Analysis and CPS Conversion for Safe, Portable, and Efficient Coroutines

Thursday January 16th, Gabriel Kerneis (Computer Laboratory - University of Cambridge) will come at IRILL to talk about "QEMU/CPC: Static Analysis and CPS Conversion for Safe, Portable, and Efficient Coroutines"

Coroutines and events are two common abstractions for writing concurrent programs. Because coroutines are often more convenient, but events more portable and efficient, it is natural to want to translate the former into the latter. CPC is such a source-to-source translator for C programs, based on a partial conversion into continuation-passing style (CPS conversion) of functions annotated as cooperative.

In this article, we study the application of the CPC translator to QEMU, an open-source machine emulator which also uses annotated coroutine functions for concurrency. We first propose a new type of annotations to identify functions which never cooperate, and we introduce CoroCheck, a tool for the static analysis and inference of cooperation annotations. Then, we improve the CPC translator, defining CPS conversion as a calling convention for the C language, with support for indirect calls to CPS-converted function through function pointers. Finally, we apply CoroCheck and CPC to QEMU (750 000 lines of C code), fixing hundreds of missing annotations and comparing performance of the translated code with existing implementations of coroutines in QEMU.

Our work shows the importance of static annotation checking to prevent actual concurrency bugs, and demonstrates that CPS conversion is a flexible, portable, and efficient compilation technique, even for very large programs written in an imperative language.

Ceylon Tour Paris 2014

January 24th, IRILL will host the Ceylon Tour Paris 2014. The Ceylon Project is a relatively new, high-level, statically and strong-typed programming language and SDK, created by Red Hat. It is based on the Java programming language.

It is a Ceylon conference, for an entire day, free, located in Paris. We will have most of the Ceylon core team, and members of the Ceylon community, who will give short talks and a workshop. Whether you don't know Ceylon yet, or want to know more, this is the place to be. We will present many aspects of the Ceylon language and ecosystem, as well as discuss the future of Ceylon.

Naturally, the conference should be a place of exchange and discussions, so we expect as much value between the talks as during the presentations.


More information on the event

Ceylon on Wikipedia

More videos on IRILL website

IRILL is glad to publish 3 more video events

Three more events recorded by IRILL are now available in video.

De l'utilisation des GPU dans la programmation Réactive/Synchrone "à la" Boussinot

Jeudi 12 décembre, Jean-Ferdinand Susini (CEDRIC-CNAM) viendra présenter ses travaux de recherche.

Le modèle de programmation Réactive/Synchrone proposé par Frédéric Boussinot a été étudié sous diverses variantes depuis le début des années 90, cependant bien que faisant la part belle à l'expression du parallélisme dans les programmes, l'essentiel des implantations (SL, SugarCubes, ReactiveML…) repose sur une traduction purement séquentielle. Quelques expérimentations ont été faites pour appuyer ce modèle sur un exécutif à base de threads (FairThread, FunLoft…), ou bien sur des systèmes répartis (Distributed Reactive Machine, Jocaml & RML…).

Le projet ANR PARTOUT a exploré différentes possibilités de prise en charge du parallélisme dans ce cadre à des niveaux de granularté très différents. Nous nous intéresserons dans cet exposé à l'intégration des calculs déportés sur GPU dans un exécutif réactif et à l'implantation qui en a été faite dans SugarCubes v5 en s'appuyant sur le modèle de calcul OpenCL.

PostgreSQL meetup at IRILL

The Parisian PostgreSQL community organizes a meetup at IRILL next Thursday

Dimitri Fontaine will present "PostgreSQL is Web Scale".

Registration on:

Debian Release team meeting

This week end (November 23 and 24), IRILL will host the Debian release team.

This internal Debian meeting will be the opportunity for the release team to prepare the next goals for Debian Jessie. The next Debian release expected for 2014/2015.

More information:

Organising data sharing, to ensure long term preservation

The November 2013 issue of the frech edition of Scientific American contains an article by Roberto Di Cosmo discussing how sharing, open standards and free software are essential for long term data preservation.

Tails: confidentialité et anonymat, pour tous et partout

L’utilisabilité et de la maintenabilité, fonctionnalités critiques pour la sécurité.

Cet exposé présentera le projet Tails ( et le système d’exploitation du même nom. Nous mettrons plus particulièrement en lumière les hypothèses spécifiques qui, selon nous, ont été décisives pour maintenir ce projet en vie plus longtemps que tout autre équivalent, et lui donner une popularité sans pareille dans sa catégorie.

30 Octobre 2013, 18h30-20h à IRILL, salle Orange, 3e étage

intrigeri est développeur Debian, et un des principaux auteur de Tails.

- Site web:

- Contact:

First Debian Meetup at IRILL

In a couple of weeks, the first Debian meetup organized by the french Debian association. It will be hosted by IRILL

L'événement aura lieu le 16 octobre 2013 à IRILL.

Logilab ( va sponsoriser les pizzas et boissons de cet évènement.


La première, classique, est le Wiki Debian:

La seconde passe par l'utilisation de


Merci d'arriver avant 19:00, il y aura quelqu'un à l'accueil. Après, il faudra que quelqu'un descende pour ouvrir.

Trois salles sont à disposition.

Nous avons prévu environ une heure de présentations et discussions dans la salle algorithme

20h00: Hello world!

20h05: Vincent Bernat - Packages de Debug

Lightning talks (environ 10 minutes) - 20 h 30 => 21 h 15

Nicolas Dandrimont - fedmsg dans Debian
Stefano Zacchiroli -
Sylvestre Ledru - Projet Debile

Suite à ces trois LT:
Discussions, trolls et questions

En parallèle, la salle protocole sera ouverte pour faire un hacklab.

Enfin, la salle de pause IRILL accueillera les logilab-pizzas.

IRILL @ Open World Forum

This week at the Open World Forum, in Paris, IRILL will be present in many different ways.

Hosted on the Inria stand, IRILL will be present to answer any of your questions.

In parallel, many members of IRILL be presenting their work:

Come to say hello!

Paris LLVM Meetup 2. October at IRILL

The LLVM meetups are back

After the success of Euro LLVM in Paris, LLVM Social in Paris is back!

On October 2nd, an LLVM Meetup will take place in Paris. We invite everybody interested in LLVM to join.

  • What is LLVM?

LLVM is a collection of modular and reusable compiler and toolchain technologies. It is used as the base of many modern compilers such as clang (C/C++/Objective-C), GHC (Haskell), dragonegg (Ada, Fortran, ...), 
several commercial OpenCL compilers or RenderScript. Industry, open source community and researchers work on or with LLVM for many reasons. 
The most cited ones are the modern code base, the liberal BSD like license and the modularity. 

Around LLVM a large ecosystem was built that includes subprojects like dragonegg, vmkit, Polly, lldb, libc++, safecode and many more.


  • What is the LLVM Meetup and Bug Squashing?

It is a casual meetup to get to know people working on LLVM or interested to learn about LLVM. There is no fixed schedule. We just meet to get to know each other and to discuss. No previous knowledge about LLVM is required. 

This time we meet at IRILL, where we will have plenty of tables, whiteboards and internet.  We plan to use this opportunity to dive into the code and to have a "Debian clang bug squashing meeting". You are invited to bring your laptop to join us (and we are happy to help with any LLVM related hacking).


  • Where/When is it?

Date:       October 2nd 2013
Time:       18:30 pm 
Location:   IRILL 
Address:    23, avenue d'Italie,75013 Paris, France 
Entering:   You can access with us at 19:00. For people who come later, 
there will be a cell phone number at the door which you can 
call to enter.


  • Do people speak English or French?

Most people attending speak both languages. It will be easy for you to communicate in your preferred language.


  • Do I need to register?

No, but if you drop us an email, we can better estimate the number of people.

Arnaud de Grandmaison 
Duncan Sands 
Sylvestre Ledru 
Tobias Grosser

Atelier Génie Logiciel Empirique

Le 23 octobre 2013 au LIP6 aura lieu le premier atelier consacré au Génie Logiciel Empirique

Le premier atelier de l'action émergente GDR-GPL "Génie Logiciel Empirique" aura lieu le 23 octobre 2013 au LIP6 (25-26/101). Au programme, un keynote de David Lo, des présentations sur les résultats les plus chauds du génie logiciel empirique en France et un brainstorm sur le présent et l'avenir de l'action et des collaborations sur le sujet.

David Lo présentera le travail suivant:

To what extent could we detect field defects? an empirical
study of false negatives in static bug finding tools.
Ferdian Thung, Lucia, David Lo, Lingxiao Jiang, Foyzur Rahman, Premkumar
T. Devanbu

Pour participer au repas et dimensionner la salle, merci de vous inscrire
via le formulaire suivant :

Videos of the GNU Hackers Meeting 2013

One week after the event, IRILL is glad to share the videos of the event:

Thanks to all the volunteers who helped on these videos.

Videos of the Debconf (Debian conference) 13

The Debconf 13 has just ended yesterday and once more, IRILL was involved in the video team of this conference.

We are happy to share the result of the work of many volunteers and speakers:

All videos are available under the Webm format and MP4 (H-264).

PRISM and an Agenda for European Network Security Research

Monday 26 August at 10, Christian Grothoff (TUM) will come at IRILL to talk about "PRISM and an Agenda for European Network Security Research"

This talk will summarize the recent revelations about PRISM and related US government espionage programs and their implications for non-American citizens, industries and governments.  We will then present technical solutions towards a secure and fully decentralized future Internet, which would address key challenges for self-determined life created by the world-wide police state.

Videos of the mini Deconf 2012 in Paris

IRILL recorded the Mini Debconf 2012 event in Paris. Since the Debconf 2013 edition just started, it is the right time to publish the videos of this event.

GNU Hackers meeting 2013 at IRILL

After the success of the 2011 edition, the GNU project and IRILL are glad to announce the 2013 edition.

After the success of the 2011 edition, the GNU project and IRILL are glad to announce the 2013 edition. During this event, hackers of the various GNU projects (gcc, guile, binutils, hurd, etc) will exchange and present their recent works and visions.

The event will take place at IRILL, 23 avenue d'Italie, Paris from Thursday 22 to the Sunday 25 of August. The talks will be recorded.

The event is open to anyone. A registration is however required.

Questions can be asked on

Official website:

Videos of the 2011 event

De l'interaction entre granularités fonctionnelle et modulaire

Jeudi 20 Juin, nous recevons Philippe Narbel (LABRI) qui nous parlera de "De l'interaction entre granularités fonctionnelle et modulaire"

Dans les architectures logicielles, la granularité modulaire est le niveau de mise en oeuvre le plus utilisé. On y relie des modules, des classes, des objets, des composants dans lesquels données et fonctions sont encapsulés.

Mais cet état de fait ne va pas sans difficultés, car la granularité fonctionnelle subsiste nécessairement, étant celle des fonctions, c'est-à-dire celle des comportements spécifiques et des calculs.

Ainsi, afin de faciliter les interactions inévitables entre ces deux granularités, il existe désormais une grande variété de mécanismes. Nous tenterons d'en présenter ici une synthèse, non seulement pour les caractériser, mais aussi pour situer ceux apparus récemment, comme par ex. les traits/mixins, les fonctions d'extension, les délégués/fermetures, les implicites, les aspects, les composants de première classe, etc.

Videos of the devops 2013 Paris edition

To provide occupation for your week end, IRILL is happy to publish the videos of the Devops Paris 2013 Edition.
To provide occupation for your week end, IRILL is happy to publish the videos of the Devops Paris 2013 Edition. 16 videos about Devops are available in both webm and mp4 (h264) formats.

Le MOOC 2013 -les vidéos

Jeudi dernier, 30 mai 2013, IRILL a coorganisé le premier événement MOOC en France.

Les vidéos de cet événement sont disponible sur le site IRILL:


Plus d'information sur l'événement:

Euro LLVM 2013 videos and feedback

Two weeks after the Euro LLVM event in Paris, IRILL is glad to share the videos of the event.

All the videos are available on the IRILL website:

A feedback of the conference has been published on the LLVM blog:

DebConf 11 and 12 videos available

To celebrate the release of Debian Wheezy (7.0), IRILL is glad to publish all the videos of the Debian Conference (DebConf) 2011 and 2012. The 128 videos are available both as OGG and MP4 (H-264) formats.

DebConf 11, the annual Debian conference took place in Banja Luka, Republika Srpska, Bosnia and Herzegovina. 56 talks or BoF have been recorded during this event:

DebConf 12 took place in Managua, Nicaragua. 72 talks or BoF have been recorded during this event, including many of them in Spanish.

Compiler Testing via a Theory of Sound Optimisations in the C11/C++11 Memory Model

Jeudi 25 avril, nous recevons CRobin Morisset - Parkas (INRA, DI ENS) avec Pankaj Pawan et Francesco Zappa Nardelli qui nous parleront de "Compiler Testing via a Theory of Sound Optimisations in the C11/C++11 Memory Model"

Compilers sometimes generate correct sequential code but break the concurrency memory model of the programming language: these subtle compiler bugs are observable only when the miscompiled functions interact with concurrent contexts, making them particularly hard to detect. In this work we design a strategy to reduce the hard problem of hunting concurrency compiler bugs to differential testing of sequential code and build a tool that puts this strategy to work. Our first contribution is a theory of sound optimisations in the C11/C++11 memory model, covering most of the optimisations we have observed in real compilers and validating the claim that common compiler optimisations are sound in the C11/C++11 memory model. Our second contribution is to show how, building on this theory, concurrency compiler bugs can be identified by comparing the memory trace of compiled code against a reference memory trace for the source code. Our tool identified several mistaken write introductions and other unexpected behaviours in the latest release of the gcc compiler.

Schedule of Euro LLVM 2013 published

In less than two weeks, the 2013 edition of the Euro LLVM conference will start in historic center of Paris, France.  The schedule has just been published.  Most of different aspects of the LLVM infrastructure will present at the event.  See below for the full details!

Monday, April 29th

Time Room Subject
11:00 - 13:00 Entrance Registration
12:30 - 13:15 ENS restaurant Lunch
13:14 - 13:30 Dussane Welcome - Day 1
13:30 - 14:30 Dussane Keynote : Optimization in LLVM - Numbers, A Case Study, and Looking Forward
14:30 - 15:15 Dussane Talk : Towards OpenMP Support in LLVM
Résistants Talk : Dagger: decompilation to LLVM IR
15:15 - 15:45 Coffee break
15:45 - 16:30 Dussane Talk : LLVM on IBM POWER processors: a progress report
Résistants Talk : Performing Source-to-Source Transformations with Clang
16:30 - 17:15 Dussane Talk : How to implement an LLVM Assembler
Résistants Talk : clang-format - Automatic formatting for C++
20:00 - 23:00 Dinner cruise on the Seine river, in Paris, for those who registered to the dinner

Tuesday, April 30th

Time Room Subject
8:45 - 9:00 Dussane Welcome - Day 2
9:00 - 10:00 Dussane Keynote : How Computers Work
10:00 - 10:45 Dussane Talk : LLVM Interpreter, a key component in validation of OpenCL compilers
Résistants Talk : Run-time tracking of uninitialized data with MemorySanitizer
10:45 - 11:05 Coffee break
11:05 - 12:05 Dussane Lightning talks
12:05 - 12:35 Résistants Posters
Cavailles Posters
12:35 - 13:35 ENS restaurant Lunch
13:35 - 14:20 Dussane Talk : lld - Solving the Linking Performance Problem
Résistants Talk : An experimental framework for Pragma Handling in Clang
14:20 - 15:05 Dussane Talk : Debug Info - Status and Directions
Résistants Talk : Implementing Data Layout Optimizations in LLVM Framework
15:05 - 16:05 Dussane Tutorial : The Clang AST - a tutorial
16:05 - 16:20 Dussane Closing word

The event is sponsored by ARM, Google, IRILL/INRIA, Intel, Parrot, QuIC, Samsung

Videos : Logiciels libres et enseignement supérieur

Les vidéos de l'événement "LSoC: Logiciels libres et enseignement supérieur" sont maintenant en ligne.

L'année dernière avait lieu la première session de la journée de conférence et discussion  Logiciels libres et enseignement supérieur.

IRILL est heureux de diffuser les 9 présentations qui ont eu lieu lors de cet événement:

Use formal methods to verify an Ada program with SPARK2014

Jeudi 11 avril, nous recevons Claire Dross de Adacore qui nous parlera de "Use formal methods to verify an Ada program with SPARK2014"

From the beginning, the Ada language was designed for critical software. It is still mostly used for the development of such systems. To facilitate the expression of requirements for the verification of programs, the last update of the Ada language includes new capabilities, such as contracts on functions or type invariants. With the increase cost of unit testing, the certification standards are opening to formal methods as a new way to address verification activities. Therefore, more and more users are interested in using formal methods for verifying part of their software.

In this talk, we present the upcoming version of the SPARK subset of Ada. It aims at easily allowing formal analysis on parts of an Ada program. We then give some hints on how formal verification is conducted in this tool. Finally, we explain how formal proof can be used in combination with more traditional testing.

« Boostez » votre code !

Inria propose une troisième édition de "Boost You Code".

Vous êtes étudiant ? Ingénieur jeune diplômé ? Le développement logiciel open source vous passionne ? Cette année encore, inria vous donne l'opportunité avec le concours Boost Your Code de présenter votre projet à un jury de scientifiques et de professionnels. Et toujours à la clé, un contrat d'un an pour développer votre projet  au sein de notre Institut.

Plus d'informations

How Firefox Makes Your JavaScript Run Faster

In the context of the Mozilla performance week, Nathan Froyd from Mozilla will explain how Firefox makes Javascript run faster.

JavaScript performance has become increasingly important over the last several years, as the size of web applications have increased and as JavaScript has become an attractive target for language translation, a sort of "bytecode for the web".

This talk will discuss several of the techniques the JavaScript engine in Firefox uses to improve JavaScript performance. We will also describe asm.js, a subset of JavaScript designed for efficient translation to machine code and suitable as a target for language translation.

On the provenance of Free and Open Source Software and the legal implications of its reuse

Jeudi 21 mars, nous recevons Daniel German qui nous parlera de "On the provenance of Free and Open Source Software and the legal implications of its reuse"

Free and Open Source Software (FOSS) has become an important source of components to reuse, in both commercial and non commercial software. While these components are free (i.e. gratis) they are made available with a license that indicates the conditions under which it can be used. As a consequence, the license of a component might restrict how it can be reused. In this talk I'll describe our recent work regarding reuse of FOSS components:

  • The problem of provenance discovery in FOSS. Components are frequently distributed in binary form, making it difficult to trace back their origin. In other cases code snippets are copied from one product to another. I'll describe Joa, a system that is capable of finding the provenance of Java classes in both binary and source code form.
  • The problem of license compliance and auditing: how to determine if a software system is satisfying all the legal constraints imposed by the components it reuses. I'll describe Ninka, a system for license identification, and a model that puts together Joa and Ninka to do license compliance of Java applications

More: Daniel's blog

Mozilla Performance Work Week at IRILL

In a week, IRILL will host the Mozilla Performance Work in Paris.

The goal is this event is to improve the performances of the various Mozilla software like Firefox or Thunderbird.

The event, internal to the Mozilla community, will take place at IRILL between the March 18 to the 22.

Patoline : la typographie "lego" en OCaml

Le 14 mars, nous recevons Pierre-Etienne Meunier du LAMA et LIAFA qui nous parlera de Patoline

Depuis TeX82, bien peu d'outils ont été développés pour la typographie automatique, en particulier des mathématiques. Dans cet exposé, je présenterai une collection d'outils et de bibliothèques OCaml, développés au LAMA en 2012, pour l'écriture de documents structurés.

L'une des idées directrices de notre approche est de rendre les parties de ce projet les plus indépendantes possible, pour éviter de tomber dans le piège de TeX et LaTeX, qui ont enfermé leurs utilisateurs dans un langage immobile pendant 30 ans. Ainsi, il est possible de remplacer des morceaux de Patoline à presque tous les niveaux, tout en restant compatible avec le reste : on peut changer le langage, l'algorithme d'optimisation, le driver de sortie, le modèle de structure des documents… tout en conservant le reste des fonctionnalités.

FOSDEM 2013 videos available

All the videos recorded during FOSDEM 2013 are now available on the IRILL website.

For example, two videos can be recommended:

Harmony : Une plate-forme d'observation de l'évolution logicielle

Xavier Blanc du Labri viendra le jeudi 28 février à IRILL pour présenter et échanger autour de Harmony.

Plusieurs métriques permettent de mesurer la qualité interne d’une application (complexité cyclomatique, profondeur de l’arbre d’héritage, dépendances d’architecture, etc.). Pour autant, le lien entre métriques et effort de développement ou de maintenance n’a pas encore été pleinement démontré. En effet, si plusieurs études montrent qu’il existe une corrélation entre métriques et maintenance, ces mêmes études montrent que les corrélations dépendent de plusieurs facteurs tels que la nature de l’application, l’organisation des développeurs mais aussi des langages de programmation utilisés.

L’équipe Génie Logiciel du LaBRI, qui effectue sa recherche sur l’évolution logicielle, a développé Harmony, un outil d’observation permettant de mesurer l’impact des métriques sur la maintenance logicielle. L’objectif étant de mesurer la valeur des métriques sur l’évolution de plusieurs applications et d’en calculer leurs impacts.

Cette présentation à l’IRILL aura pour objectif de présenter Harmony et de faire un point sur l’état de l’art des études portant sur le domaine des métriques et de l’évolution logicielle. Nous présenterons les résultats que nous avons obtenus dans ce domaine.

GSoC Meetup à IRILL

IRILL accueillera un meetup autour du Google Summer of Code (GSoC) le 13 mars prochain.

Le 13 mars prochain, IRILL accueillera un meetup autour du Google Summer of Code (GSoC).
Ce meetup est à destination de potentiels étudiants mais aussi de mentors.
Le but de ce meetup est de présenter le Google Summer of Code, son fonctionnement et ses différents jalons.
Des représentants des projets Debian, LLVM, Gnome et Scilab seront présents  lors de ce meetup pour présenter leurs projets et répondre aux questions des potentiels étudiants.

Ce meetup aura lieu le 13 mars entre 16 et 20 heures dans les locaux d'IRILL
dans la salle Algorithme.

Le sens du partage.

Le 14 février, nous recevons Thibaut Balabonski de l'équipe Gallium à Inria, qui nous parlera du sens du partage

Le partage, ou la mémorisation de résultats intermédiaires, ajoute des « raccourcis » à l'espace de réduction d'un programme fonctionnel. On pourrait penser que ces raccourcis permettent d'arriver plus vite au résultat. Ce n'est cependant qu'à moitié exact.  Nous montrerons en effet que dans le cadre de la « réduction faible », qui donne une bonne approximation des stratégies d'évaluation que l'on sait mettre en œuvre simplement, les réductions optimales avec ou sans partage ont même longueur.  En revanche, toujours en réduction faible, nous verrons que sans partage ces stratégies optimales ne peuvent pas être calculables, alors qu'avec partage il existe des stratégies optimales très simples, à commencer par « l'appel par nécessité » de Wadsworth. Le partage ne modifie donc pas la borne inférieure. Il la rend accessible.

Attention. Cet exposé contient des utilisations inusuelles du lambda-lifting et des transformations CPS.

Darcs sprint

Just like in 2011, IRILL is glad to host a sprint on Darcs, a distributed revision control system developed in Haskell.

From Friday 15h to Sunday 17th, Darcs hackers will exchange and develop on Darcs.

More information on the official website

See the 2011 edition newsreports

New video gallery on the updated IRILL website

IRILL is glad to announce the opening of its Video Gallery.

The Irill website have been undergoing some serious reconstruction, thanks to the fine people from, and the most visible consequence is the fact that we are now opening a Video Gallery, that will contain all videos recorded at events where Irill was involved.

Besides the videos of the GNU Hackers Meeting and the IRILL days videos which were already available, IRILL is proud to announce the publication of the FOSDEM 2012 edition.

Four categories are available:

More videos will be published in the next few weeks like the various Debian events, FOSDEM 2013, talks about FLOSS and education, etc.

Hosting and Automation at the OSUOSL

Le 31 Janvier nous recevons Lance Albertson, qui nous parlera de l'infrastructure utilisée à OSUSL pour héberger les plus grands projets de Logiciel Libre.

Learn how the OSUOSL operates their hosting environment which powers the core open source eco-system. Our environment is fairly unique with the variety of systems and and projects we host. We use a variety of technologies such as ganeti, kvm, glusterfs, cfengine, puppet, nagios, munin, etc. This presentation will cover what we do and how we do it. I will also cover the unique challenges we face.

Announce of the 2013 European LLVM Conference

IRILL will co-organize the third European LLVM meeting in Paris at the Ecole Normale Supérieure.

IRILL is pleased to announce the third European LLVM conference on April 29-30 2013 in Paris, France. This will be a two day conference which aims to present the latest developments in the LLVM world and help strengthen the network of LLVM developers. The format will be similar to that of the previous meetings held in London but with more time for presentations and networking. The meeting is open to anyone whether from business or academia, professional or enthusiast and is not restricted to those from Europe - attendees from all regions are welcome.

Call for Speakers, Posters, Demos

We invite academic, industrial and hobbyist speakers to present their work on developing or using LLVM, Clang, etc. Proposals for technical presentations, posters, workshops, demonstrations and BoFs are welcome. Material will be chosen to cover a broad spectrum of themes and topics at various depths, some technical deep-diving, some surface-scratching. We are looking for:

  1. Keynote speakers.
  2. Technical presentations (30 minutes plus questions and discussion) related to development of LLVM, Clang etc.
  3. Presentations relating to academic or commercial use of LLVM, Clang etc.
  4. Workshops and in-depth tutorials (1-2 hours - please specify in your submission).

The deadline for receiving submissions is March 1st, 2013. Speakers will be notified of acceptance or rejection by the 15th of March. Proposals that are not sufficiently detailed (talks lacking a comprehensive abstract for example) are likely to be rejected. Slides and posters must be in PDF format. Submissions should be done by email at Please note that presentation materials and videos for the technical sessions will be posted on after the conference.


The registration is now open on a first come, first served basis, free of charge. Please send your details (name, email, company/institution). Attendance will be limited to about 150.

A dinner will be provided on Monday evening (29th April) - this will also be free of charge. Please include whether you'll be attending the dinner on the registration form.


This meeting will be hosted by the Ecole Normale Supérieure

View Larger Map

Ecole Normale Supérieure is a university in the historic university quarter of Paris. It is easily accessible. A direct train connection (RER-B) from the train station ''Gare de Nord/Est'' as well as from Paris Charles de Gaulles airport leaves you at RER station ''Luxembourg'' which is in walking distance to Ecole Normale Supérieure. The metro stations "Cardinal Lemoine", "Place Monge" and "Censier-Daubenton" are also nearby.

Metro Map

Access Map


The mailing list can be used to discuss issues of general interest related to the conference organization, such as suggestions for special kinds of side-sessions.

Financial Support

There may be a possibility of limited funding to help students or contributors who could not otherwise attend the conference. This will depend on overall sponsorship and companies' interest in supporting the event. If you need funding to attend the meeting, or can help sponsor, please tell us on the registration form.

Managing Heterogeneity and Variability in Cloud Computing Environments

Le 23 janvier à 10h30, Clement Quinton (INRIA) viendra présenter son travail autour de "Managing Heterogeneity and Variability in Cloud Computing Environments".

Cloud Computing is a major trend in distributed computing environments enabling software virtualization on configurable runtime platforms. Configuration and customization choices arise due to the heterogeneous and scalable aspect of the cloud computing paradigm. The wide range of cloud solutions and the lack of knowledge in this domain is a real problem for companies when facing the cloud solution choice. Indeed, the selection of a cloud among others as well as the selection of required tools and/or libraries in this cloud remains challenging due to the amount of providers and their intrinsic variability. Feature Models (FM) originating from Software Product Lines (SPL) approach is one way to handle this variability and then manage and create configurations in the cloud, while ontologies are used to model the various semantics of cloud systems. We propose a model-driven approach based on FMs and ontologies. We use extended FMs to represent cloud providers variability and their resources dimensions, combined with ontologies that capture cloud computing knowledge. We define a mapping between ontologies and FMs to bridge the gap between the application configuration and cloud platforms’ offer.

Une définition CPS d'Esterel par Bernard Serpette

Mardi 18 décembre à 13h, Bernard Serpette (Indes - Inria Sophia Antipolis) présentera son travail autour d'une définition CPS d'Esterel

Dans le cadre du langage d'orchestration HipHop, extension de Hop, le projet Indes a été amené à définir un interprète "à la Esterel". Le problème principal consiste à définir un prédicat assurant l'absence d'un signal à un certain instant. Pour ce faire nous élaborerons une analyse statique, calculée à la compilation, permettant d'obtenir une estimation des émissions de signaux à venir.

De plus cette définition d'Esterel basée sur les continuations permet d'obtenir un interprète n'allouant pas de mémoire pour ses besoins propre : les continuations peuvent être pré-allouées au moment de la compilation.

CamllVM - a LLVM based runtime for Ocaml

Le 20 décembre à 15h45, Raphael Amiard (UPMC) viendra présenter son travail autour de CamllVM - a LLVM based runtime for Ocaml.

There are two official ways of running Ocaml code : The first is using the bytecode compiler and interpreter, the second is using the ocamlopt compiler to generate native code directly.

There has been many discussions about using LLVM in the context of OCaml code generation. The theoretical advantages of using LLVM for code generations are amongst others, faster execution, better portability, and shared optimizations amongst all users of LLVM.

The CamllVM project takes the approach of generating LLVM code from the OCaml bytecode usually interpreted by ocamlrun. There are many problems to tackle in such an approach, but the two main pain points arise from two facts :

  • OCaml bytecode was meant to be interpreted, not compiled. It is untyped, where LLVM intermediate representation is typed, and uses stack semantics where LLVM uses register based semantics.
  • LLVM, despite aiming to be a general library apt to compile down all sorts of languages, is of course designed with C, C++, and Objective-C in mind, since it is the main languages it is used to compile. It thus presents a number of difficulties to us, mainly about handling exceptions and garbage collection.


In this talk, I will present the advantages and drawbacks of the approach CamllVM has taken, what leads were explored, and what is working. I will also summarize the main difficulties in compiling OCaml to LLVM.

Experiments in generic programming: runtime type representation and implicit values by Gregoire Henry

Le 22 novembre à 15h45, Grégoire Henry viendra présenter son travail sur "Experiments in generic programming: runtime type representation and implicit values" à place d'Italie.


We present two ongoing related experiments in generic programming with OCaml. In the first experiment, inspired from the implicit values of Scala, the language is extended with primitives to enrich a locally scoped environment of implicit values -- mapping types to values -- and a construcion for synthesising a value of a given type from the environment of implicit values. We have used this extension to implement light-weight, semi-implicit type classes.

The second experiment is based on explicit runtime-type representation, described by a GADT, that can be analysed in a type-safe manner to implement polytypic functions, defined by cases on the structure of the types of their arguments.

Both experiments share a common set of non-intrusive extensions of the OCaml compiler; in particular, the modication in the type checker are simple, limited, and identical in both cases. This is because implicits can be implemented modularly as an additional phase in the compilation chain that comes right after type-checking.

Speaker: Gregoire Henry (PPS)

Proving cryptographic schemes in Agda by Nicolas Pouillard

Le 14 novembre à 15h45, Nicolas Pouillard viendra présenter son travail sur "Proving cryptographic schemes in Agda: Dependently typed functional programming for Alice and Bob" sur le site Jussieu.


In order to gain confidence in cryptographic schemes and primitive operations, cryptographers use a combination of theories. These theories enable us to mathematically prove security properties of these constructions. Probability theory, game theory, number theory and complexity theory are often all needed at the same time. We aim at realizing these proof arguments in Type Theory using Agda. Our current approach is to carefully use abstractions and dependently typed functional programming techniques to elegantly prove security properties. Finally we aim at reducing the usual gap between the formal description of the construction and its actual implementation. We do so by writing in a functional style supporting extraction into low level circuits.


Nicolas Pouillard - ITU (IT University of Copenhagen), DemTech project


Site Jussieu, salle 332 couloir 26-00, au 3ème étage

JFLO’12: Journée Francophone de Leçons sur Ocsigen

Le 14 décembre prochain, à destination des programmeurs OCaml, une journée didactique autour de Ocsigen est organisée.

IRILL organise une journée didactique autour d'Ocsigen dédiée aux programmeurs OCaml. Cette journée aura lieu le 14 décembre à l'IRILL (cf infos pratiques plus bas). Vous êtes invités à vous inscrire à l'adresse suivante:

Au programme, des cours et tutoriels sur les principaux composants du projet Ocsigen :
- Programmation concurrente avec Lwt.
- Programmation côté client avec le compilateur Js_of_ocaml.
- Programmation d'applications client-server avec Eliom.
Informations pratiques :
LIEU : Salle ORANGE, 5e étage, Inria, 23, avenue d'Italie,75013 Paris
DATE : Vendredi 14 Décembre de 10h00 à 18h00.
Informations importantes :
- les cours seront en français.
- il est fortement conseillé de venir muni de son ordinateur portable
- si possible avec une version récente d'Eliom installée

Une introduction informelle aux méthodes formelles par Prof. Roberto Bagnara, Université de Parma

Le 21 novembre, Prof. Roberto Bagnara de l'Université de Parma viendra présenter une introduction informelle aux méthodes formelles à IRILL

La vérification et la validation des systèmes logiciel critiques est une tâche ardue, pour laquelle les méthodes formelles sont de plus en plus recommandés par des standards industriel de référence, comme le DO-178C (avionique) et l'ISO 26262 (automobile).

Dans cet exposé, qui a été conçu spécifiquement pour une audience industrielle, nous présentons une introduction à trois classes de méthodes, adaptées à la spécification, l'analyse et la vérification de propriétés: les approches déductives, le model checking et l'interprétation abstraite.

On s'appuyera dans le cours de l'exposé sur des exemples et des applications qui permettront de saisir plus facilement les concepts fondamentaux de chaque classe de méthodes.

Presentation of Jesus M. Gonzalez-Barahona about "Measuring free software development"

The October 11th at 3:45, Jesus M. Gonzalez-Barahona will present his work about "Measuring free software development" at IRILL.

 Most free / open source software projects have publicly available repositories with many details about how their activity.

All changes are recorded in the source code management system (Subversion, git, Mercurial, bazaar, etc.), along with who and when made the change, and other metainformation. Issue tracking (ticketing) systems (Bugzilla, Jira, etc.) record not only bug reports and how they are fixed, but also feature requests, design discussions, etc. Mailing lists, irc logs or forums carry information about discussions, user support, etc. All of this information, when properly analyzed, can be used to track how a project is performing, and to detect interesting patterns and potential or real problems early.

The talk will present how all if this can be done with the tools in the MetricsGrimoire toolset, including some examples for specific projects.

Structured approaches for multi/many core targeting by Marco Danelutto

Marco Danelutto (DI - Università di Pisa) will be presenting "Structured approaches for multi/many core targeting" at IRILL September 18th at 16:30

  The advent of multi/many cores is changing the scenario where parallel programming has to be used. We present some recent advances relative to the programming models and tools suitable to target shared memory multicores as well as many cores (GP-GPUs but also GP many cores such as Tilera Pro64). In particular, we discuss results relative to FastFlow ( and to macro data flow techniques supporting structured parallel programming.

IRILL hosts the next FusionForge meeting

The FusionForge project developers will gather on October 10th, 2012 for a meeting hosted on the premises of IRILL in Paris.

IRILL hosts the next FusionForge meetingFusionForge runs several key collaboration platforms (forges) used notably to host various Free Software development projects including Debian, Blender, OCaml and many others hosted by such entities as INRIA and Adullact.

The community of developers works online (using FusionForge) and usually coordinates through IRC meetings and mailing lists, but once in a while, physical meetings are necessary to discuss the overall progress of the projects, challenges and hack on technical problems.

FusionForge 5.2 should be released mid-september, and the meeting should happen in a good timing for the project to draft its future roadmap. Users and developers of other forge engines are welcome to attend and exchange ideas.

The last meeting occurred in 2010 in Paris, and the project will gather again in Paris at IRILL, on the eve of the Open World Forum.

Bootstrapping Debian (or derivatives) for a new architecture by Johannes Schauer

Le 20 septembre à 15h30, Johannes Schauer viendra présenter son travail sur "le Bootstrapping de Debian sur des nouvelles architectures" chez IRILL.

Bootstrapping Debian for a new architecture has so far been a manual task often involving months of work because neither cross compilation nor compilation of source packages with reduced dependencies is supported by Debian. Packages have to be cross compiled to provide a minimal native compilation environment. Packages have to be built with reduced dependencies to break build dependency cycles. Using multiarch, Debian will soon support cross compilers and buildprofiles will allow to build a source package with less build dependencies.

This project contains tools for three purposes:
Firstly, finding out which packages have to be cross compiled for a minimal system.
Secondly, which packages should be considered for reduced build dependencies or cross compilation by finding elementary circuits in the build dependency graph.
Thirdly, what is the final order in which packages have to be cross compiled, stage compiled or normally compiled to bootstrap all of Debian.

The talk will cover the developed techniques, the findings made and the open problems remaining.


Bachelor of Computer Science at Jacobs University Bremen, currently in Master studies. Bachelor Thesis about network flow trace analysis. Current research projects about approximate solutions for the 3D orthogonal knapsack problem and about integrating RPL and SNMP on contiki OS.

Paris LLVM Meetup 25. September at IRILL

For the fourth time, a LLVM meetup is organized in Paris.

  On 25 september, an LLVM Meetup will take place in Paris. We invite everybody interested in LLVM to join. 

  • What is LLVM? 

LLVM is a collection of modular and reusable compiler and toolchain technologies. It is used as the base of many modern compilers such as clang (C/C++/Objective-C), GHC (Haskell), dragonegg (Ada, Fortran, ...), 
several commercial OpenCL compilers or RenderScript. Industry, open source community and researchers work on or with LLVM for many reasons. 
The most cited ones are the modern code base, the liberal BSD like license and the modularity. 

Around LLVM a large ecosystem was built that includes subprojects like dragonegg, vmkit, Polly, libc++, safecode and many more. 


  • What is the LLVM Meetup and Bug Squashing? 

It is a casual meetup to get to know people working on LLVM or interested to learn about LLVM. There is no fixed schedule. We just meet to get to know each other and to discuss. No previous knowledge about LLVM is required. 

This time we meet at IRILL, where we will have plenty of tables, whiteboards and internet.  We plan to use this opportunity to dive into the code and to have a "Debian clang bug squashing meeting". You are invited to bring your laptop to join us (and we are happy to help with any LLVM related hacking). 


  • Where/When is it? 

Date:       September 25 2012 
Time:       19:00 pm 
Location:   IRILL 
Address:    23, avenue d'Italie,75013 Paris, France 
Entering:   You can access with us at 19:00. For people who come later, 
             there will be a cell phone number at the door which you can 
             call to enter. 


  • Do people speak English or French? 

Most people attending speak both languages. It will be easy for you to communicate in your preferred language. 


  • Do I need to register? 

No, but, if you drop us an email at, we can better estimate the number of people. 

Arnaud de Grandmaison 
Duncan Sands 
Sylvestre Ledru 
Tobias Grosser 

Puissance de Clojure par Tom Marble

Mardi prochain, le 26 juin à 15h45, Tom Marble viendra présenter Clojure

Cette présentation et discussion éxplorera les avantages Clojure - un langage né de Java et Common LISP sans leurs inconvénients. Malgré l'absence d'une norme et avec q'une implementation Clojure bénéficie d'une communauté actif et de l'innovation continue.

 L'approche de Clojure est de la programmation fonctionnelle. Le langage évite les effets secondaires en favorisant l'immutation par défaut. Ceci permet un modèle facile pour la programmation concurrente grace au système de mémoire transactionnelle et les agents asynchrones.

 La polymorphisme dynamique libère les développeurs du «royaume des noms» -- c'est à dire un graphe simple des types orientée objets. Les multimethods de Clojure assurent la sûreté de l'abstraction des types.



Mr. Marble is best known for being on the core team at Sun Microsystems that open sourced Java.

Most recently he has launched Informatique, Inc.: a consultancy to leverage his electrical engineering background along with his software development experience for clients in such domains as telematics for electric vehicles, smart grid interoperability, probabalistic model checking, autonomous cyber defense, multiplayer online gaming and highly concurrent software design.

 Mr. Marble is an enthusiastic open source software contributor and specializes in Clojure and the Debian GNU/Linux distribution. He has been a speaker at JavaOne, Debconf, UbuntuLive, OSCON, FISL and FOSDEM.



Dart: Un nouveau langage de programmation pour le web de Nicolas Geoffray (Google)

Jeudi prochain, le 21 juin à 15h45, Nicolas Geoffray (Google) viendra présenter son travail sur "Dart: Un nouveau langage de programmation pour le web" chez IRILL.

Dart est un nouveau langage de programmation pour la création d'applications web, conçu dans un souci de performance et de simplicité. Le langage est une combinaison de syntaxe similaire à d'autres langages de programmation, et de fonctionalités moins communes telles que les isolates, ou le typage optionnel. Lors de cette présentation, j'exposerai les raisons de la création d'un nouveau language, à quoi il ressemble et les outils qui l'entourent, en espérant vous donner une bonne idée de ce qu'il peut vous apporter dans la creation de larges applications webs.

Paris LLVM Meetup 26. June at IRILL (with clang bug squashing)

For the second time, a LLVM meetup is organized in Paris. This meetup is joint with a clang bug squashing party.

 On Tuesday, 26. June, an LLVM Meetup will take place in Paris. We invite everybody interested in LLVM to join. 

  • What is LLVM? 

LLVM is a collection of modular and reusable compiler and toolchain technologies. It is used as the base of many modern compilers such as clang (C/C++/Objective-C), GHC (Haskell), dragonegg (Ada, Fortran, ...), 
several commercial OpenCL compilers or RenderScript. Industry, open source community and researchers work on or with LLVM for many reasons. 
The most cited ones are the modern code base, the liberal BSD like license and the modularity. 

Around LLVM a large ecosystem was built that includes subprojects like dragonegg, vmkit, Polly, libc++, safecode and many more. 


  • What is the LLVM Meetup and Bug Squashing? 

It is a casual meetup to get to know people working on LLVM or interested to learn about LLVM. There is no fixed schedule. We just meet to get to know each other and to discuss. No previous knowledge about LLVM is required. 

This time we meet at IRILL, where we will have plenty of tables, whiteboards and internet.  We plan to use this opportunity to dive into the code and to have a "Debian clang bug squashing meeting". You are invited to bring your laptop to join us (and we are happy to help with any LLVM related hacking). 


  • Where/When is it? 

Date:       26. June 2012 
Time:       19:00 pm 
Location:   IRILL 
Address:    23, avenue d'Italie,75013 Paris, France 
Entering:   You can access with us at 19:00. For people who come later, 
             there will be a cell phone number at the door which you can 
             call to enter. 


  • Do people speak English or French? 

Most people attending speak both languages. It will be easy for you to communicate in your preferred language. 


  • Do I need to register? 

No, but, if you drop us an email at, we can better estimate the number of people. 

Arnaud de Grandmaison 
Duncan Sands 
Sylvestre Ledru 
Tobias Grosser 

Software Carpentry

End of june, the Software Carpentry event will be hosted at Inria and co-organized by IRILL. The ggoal is to help scientists and engineers become more productive by teaching them basic computing skills like program design, version control.

 When: June 28-29, 2012.

Where: INRIA Paris (Place d’Italie)
What: Our goal is to help scientists and engineers become more productive by teaching them basic computing skills like program design, version control. In this two-day boot camp, short tutorials will alternate with hands-on practical exercises. Participants will be encouraged both to help one another, and to apply what they have learned to their own research problems during and between sessions.
Who: The course is aimed at postgraduate students and other scientists who are familiar with basic programming concepts (like loops, conditionals, arrays, and functions) but need help to translate this knowledge into practical tools to help them work more productively.
Requirements: Participants must bring a laptop with a few specific software packages installed. (The list will be sent to participants a week before the boot camp.)
Content: The syllabus for this boot camp will include:
  • Using version control to manage and share information: an introduction to git and github
  • Introduction to Python for scientists
  • Numerical programming with Numpy and Scipy


Iterative Delta Debugging: Analyzing the evolution of components

Jeudi prochain, le 31 mai à 15h45, Cyrille Artho (AIST) viendra présenter son travail sur "Iterative Delta Debugging: Analyzing the evolution of components" chez IRILL.

Automated debugging attempts to locate the reason for a failure. Delta debugging minimizes the difference between two inputs, where one input is processed correctly while the other input causes a failure, using a series of test runs to determine the outcome of applied changes. Delta debugging is applicable to inputs or to the program itself, as long as a correct version of the program exists. However, complex errors are often masked by other program defects, making it impossible to obtain a correct version of the program through delta debugging in such cases. Iterative delta debugging extends delta debugging and removes series of defects step by step, until the final unresolved defect alone is isolated. The method is automated and managed to localize a bug in some real-life examples.

We believe that this idea of analyzing software changes over time can also be applied to the evolution of software components/packages, and help with finding conflict defects between multiple software packages.

Programme des journées sur la Fiabilité du Logiciel


Journées IRILL / GTLL


sur la Fiabilité du Logiciel


14-15 Juin 2012,


Salle Orange


23 Avenue d'Italie - 5th floor


Paris, France 


La participation est gratuite, mais les places sont limitées: inscrivez vous dès maintenant!


Retrouvez des informations sur comment venir.



Motivation et Objectif des journées

Le logiciel est devenu pervasif: on le retrouve désormais dans tous les processus industriels, administratifs et individuels, il est présent dans toute sorte de dispositif électronique, et il voit se généraliser l'usage de composants logiciels libres dans toute sorte d'applications.

Avec la généralisation de ces usages, la fiabilité du logiciel est une propriété qui devient de plus en plus essentielle,et il est important de favoriser les échanges entre académiques et industriels sur ces sujets.

Pour contribuer à cela, l'IRILL et le GTLL de Systematic organisent deux journées dédiées aux académiques et industriels de la région Ile de France sur la validation, la certification, la preuve et l'analyse de code.

Une série de courtes présentations permettront de découvrir une sélection d'outils de pointe, issus de la recherche, et des retours d'expérience industriels.

Thursday 14th June 2012


9h-9h30 Registration and breakfast


9h30-9h45 Welcome address from the organisation committee (Emmanuel Chailloux, Roberto Di Cosmo, Ralf Treinen)


9h45-10h20 Frama-C (Benjamin Monate)

10h20-10h55 Why (Jean-Christophe Filliâtre)

10h55-11h00 Welcome from Inria's direction (Isabelle Ryl, Antoine Petit)

11h00-11h25 Coffee Break

11h25-11h50 Compcert (Xavier Leroy)

11h50-12h25 Adacore (Yannick Moy)


12h30-14h00 Lunch break


14h00-14h45 Celia (Mihaela Sighireanu)

14h45-15h20 Malware detection (Tayssir Touili)

15h20-15h50 Coffee Break

15h50-16h25 CosyVerif (Fabrice Kordon)

16h25-17h00 TBA

End of day one


Friday, June 15th 2012


9h-9h30 Breakfast


9h45-10h20 Fluctuat (Eric Goubault)

10h20-10h55 RATP (Evguenia Dmitrieva)

10h55-11h25 Coffee Break

11h25-11h50 Imitator (Laurent Fribourg)

11h50-12h25 Clearsy (Thierry Lecomte)



12h30-14h00 Lunch break


14h00-14h45 EADS (Louis Granboulan)

14h45-15h20 Coccinelle (Gilles Muller)

15h20-15h50 Coffee Break

15h50-16h25 Trusted Labs (Boutheina Chetali)

16h25-16h40 Clôture.

End of the event.


LLVM Social in Paris

On Tuesday, 15. Mai, an LLVM Social will take place in Paris. We would like to invite everybody interested in LLVM to join us.

 o What is LLVM?

LLVM is a collection of modular and reusable compiler and toolchain technologies. It is used as the base of many modern compilers including clang (C/C++/Objective-C), GHC (Haskell), dragonegg (Ada, Fortran, ...),
many commercial OpenCL compilers, RenderScript and many more. Industry, open source community and researchers work on or with LLVM for many reasons. The most cited ones are the modern code base, the liberal BSD like license and the modularity. All making it easy to use LLVM.

LLVM in the context of this social means everything related to LLVM. This includes the core LLVM, but also projects developed in the context of LLVM. These are e.g. vmkit developed at IRILL, Polly developed at ENS, but also lldb (the LLVM debugger), libc++ (a C++ standard library) or in general projects only interested in using LLVM.

o What is a Social?

It is a casual meeting to get to know people working on LLVM or interested to learn about LLVM. There is no fixed schedule and no specific topic. We just meet to have a some drinks to get to know each other. To attend, no previous knowledge about LLVM is required.

o Where/When is it?

Date:       15. Mai 2012
Time:       19:00 pm
Location:   Brasserie - Le Lakanal
Adress:     2, Avenue Rene Coty, 75014 Paris
Metro-RER:  Denfert-Rochereau

o Do people speak English or French?

Many people attending speak both languages. It will be easy for you to
communicate in your preferred language.

o Do I need to register?

You can just pass by, but it would be nice if you drop us a line at, such that we can reserve some tables. 



La participation aux journées est gratuite, mais les places sont limitées! Inscrivez vous ici: Loading...

Access to Irill machines from outside

You need to have your public ssh key installed on the "bastion ssh": instructions (as for 2012/05) here:

WARNING: you can only perform the procedure while you are on a physical Inria network.

Haxe, le langage multi-plateforme basé sur OCaml

Jeudi 3 Mai 2012, Alexandre Hoyau (Silex Labs)  presentera Haxe, un langage multi-plateforme basé sur OCaml.

Tooling for open source software license compliance

Jeudi prochain, le 5 avril à 15h45, Armijn Hemel (Tjaldur Software Governance Solutions) viendra présenter son travail sur "Tooling for open source software license compliance - challenges in fixing trivial problems" chez IRILL.

Open Source software is everwhere. Its open nature and easy availability has significantly stimulated software reuse in every industry, ranging from tax software to avionics. Unfortunately the license terms are often not very well understood or even ignored, which can put companies at significant risk. This is especially visible in consumer electronics, where there have been several lawsuits in Germany, France and the USA.

These lawsuits were, from an engineering point of view, completely useless: proper processes and best practices that have been known for decades in software engineering would have license compliance a trivial problem. Reality is a bit different: companies often don't know what is inside software that they buy, how it was made and what license the software is under. These companies often don't have the budget to invest in expensive tooling, that often does not even solve their problem namely finding out what is in the binaries they are shipping. There is a huge need for cheap, easy and reliable tooling.

In this talk I will briefly touch upon backgrounds about why license violations happen, where problems occur and where they should be fixed, plus introduce the Binary Analysis Tool, an Apache 2 licensed tool for analysing abritrary binary files, such as firmwares, executables and file systems.

Biography: Armijn Hemel, MSc is the General Manager of Tjaldur Software Governance Solutions, a company specializing in offering tooling, training and consultancy in the field of open source software license compliance and software governance. He has been active in the project since mid 2005, where he has helped resolve several hundred license violations.

Voir aussi:

Tjaldur Software Governance Solutions


Pattern matching for the masses using custom notations par Nic Volanschi (Matchbox & Metaware)

Jeudi prochain, le 22 mars à 15h45, Nic Volanschi (Matchbox & Metaware) viendra présenter son travail sur "Pattern matching for the masses using custom notations" chez IRILL.

Le filtrage (matching) utilisant des motifs déclaratifs (patterns) a fait ses preuves dans de nombreux langages fonctionnels et logiques, en montrant qu'il permet souvent d'écrire des programmes plus concis, plus élégants, et plus lisibles. Malgré ce succès de longue date, le filtrage n'a pas encore trouvé sa voie vers la grande majorité des programmeurs, car il a toujours été négligé par les concepteurs de langages populaires tels que Java, C#, ou JavaScript. Nous montrons que le filtrage par motifs, y compris utilisant des notations personnelles, est facile à integer dans différents langages à objets sous la forme d'une simple bibliothèque, et nous illustrons cette démarche à travers un prototype ouvert nommé Matchbox.

L'exposé sera focalisé sur l'applicabilité pratique de cette démarche, en illustrant aussi ses limitations et des pistes pour les contourner. Enfin, nous esquisserons l'ouverture d'un nouvel espace d'innovation ouverte dans lequel tout programmeur peut sortir de son rôle traditionnel de consommateur de langages pour en devenir un contributeur.

Voir aussi: Metaware

Verifying Fence Elimination Optimisations par Francesco Zappa Nardelli (projet Moscova - Inria)

Jeudi prochain, le 8 mars à 15h45, Francesco Zappa Nardelli (projet Moscova - Inria) viendra présenter son travail sur "Verifying Fence Elimination Optimisations" chez IRILL.


We consider simple compiler optimisations for removing redundant memory fences in programs running on top of the x86-TSO relaxed memory model. While the optimisations are performed using standard thread-local control flow analyses, their correctness is subtle and relies on a non-standard global simulation argument. The implementation and the proof of correctness are programmed in Coq as part of CompCertTSO, a fully-fledged certified compiler from a concurrent extension of a C-like language to x86 assembler, built on top of CompCert. In this talk, after recalling the x86-TSO model, I will describe the optimisations and their soundness proof.

VMKit: a Substrate for Managed Runtime Environments

Jeudi prochain, le 23 février à 15h15, Gaël Thomas (REGAL - LIP6 - UPMC) viendra présenter son travail autour de VMKIT chez IRILL.

Managed Runtime Environments (MREs), such as the JVM and the CLI, form an attractive environment for program execution, by providing portability and safety, via the use of a bytecode language and automatic memory management, as well as good performance, via just-in-time (JIT) compilation. Nevertheless, developing a fully featured MRE, including e.g. a garbage collector and JIT compiler, is a herculean task. As a result, new languages cannot easily take advantage of the benefits of MREs, and it is difficult to experiment with extensions of existing MRE based languages.

This talk will describe and evaluates VMKit, a first attempt to build a common substrate that eases the development of high-level MREs. We have successfully used VMKit to build two MREs: a Java Virtual Machine and a Common Language Runtime. An extensive study of this infrastructure assesses the ease of implementing new MREs or MRE extensions and evaluates the resulting performance. In particular, it took one person only one month to develop a Common Language Runtime using VMKit. VMKit furthermore has performance comparable to the well established open source MREs Cacao, Apache Harmony and Mono, and is 1.2 to 3 times slower than JikesRVM on most of the DaCapo benchmarks.

Conférence de Richard Stallman "Pour une société numérique libre"

Le jeudi 16 février à 18h30, à l'université Paris-Diderot, Richard Stallman, fondateur du mouvement du logiciel libre,donnera une conférence exceptionnnelle intitulée "Pour une société numérique libre".


"Les efforts développés pour inciter les gens à utiliser les nouvelles technologies numériques sont réalisés sur la supposition que cette utilisation est invariablement une bonne chose. En jugeant par le seul aspect pratique immédiat, cela semble être le cas. Cependant, si on juge en termes de droits de l'homme, que cette entrée dans ce monde soit bonne ou mauvaise dépend du monde numérique dans lequel nous voulons être inclus. Si nous nous fixons cette inclusion comme objectif, il nous incombe de nous assurer que cela soit réalisé de la bonne manière."

La conférence sera réalisée en français et ne concernera pas d'aspects techniques.


Elle se tiendra à l'amphi Buffon de l'université Paris-Diderot, le jeudi 16 février, à 18h30.
Elle est organisée par l'IRILL et l'Université Paris-Diderot.
Elle sera enregistrée et disponible par la suite.
Les portes seront ouvertes à 18h10. En cas de forte affluence, un second amphithéâtre voisin sera disponible avec la retransmission depuis l'amphi Buffon.
Adresse : 13 rue Hélène Brion, 75013 Paris
Métro et RER : station Bibliothèque François-Mitterand, ligne 14 et RER C.

Plan :

Rester en contact avec la FSF :

Université Paris-Diderot :

Vidéos de l'évènement

Architecture de Data Publica (Groupe de travail "Programmation")

Le 9 février à 15h45, Christian Frisch de Data Publica ( viendra présenter leur travail autour de l'importation de données publiques (ou privées).

Titre : Architecture de Data Publica

Auteur : Christian Frisch (Data Publica)

Abstract : Data Publica répertorie et indexe les données françaises, qu'elles soient publiques ou privées, gratuites ou payantes, locales/nationales/internationales. Les jeux de données (en général des fichiers xls) sont repertoriés initialement manuellement, maintenant automatiquement au moyen de technologies de crawl/scraping. Ils sont ensuite indexés (metadonnées et full texte) dans un moteur de recherche (Solr) et des prévisualisation html sont générées (openoffice-server). Nous avons notamment importé en "bulk mode" les 5000 jeux de données d'Eurostats et les 2700 de

Plus récement, nous avons mis en place une base de données pour stocker une version structurée de ces données. Cette base (MongoDB) utilise un modèle conceptuel générique adapté  la donnée publique (Google's DSPL). Cette modelisation permet d'acceder aux données sous forme d'une API générique (type REST/Json) et d'un visualisateur générique (GWT/highcharts/maps). Pour alimenter cette base, nous créons des jeux de données au format DSPL:

  • manuellement
  • à partir d'un outil de reconnaissance de tableau développé par INRIA/Zenith
  • en transformant des jeux de données déjç  formatté (type SDMX)

Enjeux actuels:

  • Chargement en masse de jeux de données à  partir des jeux Eurostat, Banque Mondiale, OCDE... et le enjeux de passage à l'echelle
  • "sémantiser les jeux de données" c'est à dire lier nos concept aux entités correspondantes du monde sémantique (DBPedia, FreeBase, GeoNames...). Construction d'un moteur de recherche qui prend en compte cette modélisation.

Install party à l'ENS-Cachan organisé par Cr@ns

Le Cr@ns, l'association gérant le réseau des élèves à l'ENS-Cachan, organise une install-party le Samedi 14 Janvier. Durant l'après midi, plusieurs acteurs du monde du logiciel libre en présenteront différent aspects. L'IRILL sera présent et fournira l'infrastructure pour filmer ces conférences.

14h - 15h : Luca Saiu

"Le mouvement du logiciel libre"

Le mouvement du logiciel libre est basé sur quatre libertés : la liberté d'utiliser un programme pour n'importe quelle finalité, la liberté de le modifier, celle d'en redistribuer des copies exactes, et celle d'en distribuer des copies modifiées qui contribueront à son développement. Né en 1983 de l'«idéalisme pragmatique» d'une seule personne essayant de reconstruire une communauté centrée sur les droits des utilisateurs, le logiciel libre profite aujourd'hui d'un développement exceptionnel en quantité et qualité des productions. Nous allons parler de l'histoire et de la philosophie de ce mouvement fascinant, des aspects légaux et pratiques de l'utilisation du logiciel libre et de son développement.

15h - 16h : Stéphane Bortzmeyer

"Peut-on se passer de moteurs de recherche ?"

Aujourd'hui, beaucoup d'utilisateurs dépendent entièrement d'un moteur de recherche pour toute navigation sur l'Internet. On entend même des enseignants de collège dire aux élèves « Pour aller sur Wikipédia,
tapez "wikipedia" dans Google » Pourquoi est-ce une mauvaise idée ?
Quels sont les inconvénients des moteurs de recherche ? Que se passe-t-il lorsqu'une panne ou la censure modifie le résultat d'une recherche ? Quels sont les points forts des moteurs de recherche, où ils sont utiles ? À quoi
sert le DNS et pourquoi est-ce important de comprendre les noms de domaine ?

Une amusante vidéo qui illustre bien les inconvénients de la dépendance envers Google :

16h - 17h : Jean-Vincent Loddo

"Tirer les ficelles de l'architecture TCP/IP avec Marionnet"

Marionnet est un logiciel permettant de définir, configurer, exécuter et contrôler un réseau virtuel constitué d'ordinateurs utilisant le système d'exploitation GNU/Linux, de concentrateurs, de commutateurs,
et de routeurs. Avec Marionnet, il est possible d'expérimenter la mise en oeuvre complète d'un réseau local : le projet, le câblage, le lancement, la configuration, l'administration, l'étude de protocoles et le test de
services ou d'applications. Il permet de pratiquer, analyser et contrôler les différentes couches réseaux qui constituent l'architecture DoD (TCP/IP): depuis le niveau physique jusqu'au niveau application. Adopté dans une
structure universitaire, Marionnet permet de réduire l'utilisation des vraies salles de « TP réseau », constituées d'équipements souvent onéreux et difficiles à maintenir en parfait état de marche. Il permet aussi aux étudiants de travailler à distance ou dans les salles informatiques ordinaires en accès libre.

Planning du LSoC: Logiciels libres et enseignement supérieur

Comme indiqué dans un message précédent, IRILL organise et accueillera  une journée de conférences et de discussions sur l'enseignement de technologies des logiciels libres dans l'enseignement supérieur. L'organisation de la journée est maintenant connue:

9h   - 9h30     : Réception, café
9h30 - 10h      : Session d'ouverture, Enseignement supérieur et logiciels libres, panorama des initiatives en France. (Ivaylo Ganchev, université Paris 8)
10h  - 10h30    : Les initiatives autour du libre à l'EPITA (Olivier Ricou, EPITA)
10h30  - 11h    : L'enseignement des technologies Mozilla à l'université d'Evry - CoMETE (Didier Courtaud, CEA)

11h   - 11h20   : Pause

11h20 - 11h50   : Développement de VLC à l'Ecole centrale de Paris (Jean-Baptiste Kempf, Videolan)
11h50 - 12h20   : Réalisation de projets libres dans le cadre d'un cours magistral à l'université Paris 8 (Karine Mordal, université Paris 8)

12h20 - 14h     : Pause déjeuner

14h     - 14h30 : Développement de module Open Office à l'école centrale de Nantes (Morgan Magnin, Nelle Varoquaux, Benjamin Vialle, EC Nantes)
14h30 - 15h     : Enseignement des technologies KDE à l'université Paul Sabatier Toulouse (Kevin Ottens, KDE)
15h     - 15h30 : Utilisation du logiciel libre à l'université du Littoral: Master logiciel libre (Cyril Fonlupt, Eric Ramat, Master I2L, université du Littoral)

15h30  - 16h    : Pause

16h - 16h15 : La nuit de l'info - retour sur expérience (Pierre-Allain Muller, Université de Haute-Alsace)
16h15     - 17h30 : Table ronde : Libre semester of Code, mutualisation de ressources pédagogiques et autres sujets discutés pendant la journée (Albert Cohen, Roberto Di Cosmo)


LSoC: Logiciels libres et enseignement supérieur

                                            Inscription gratuite, mais obligatoire, sur ce formlaire en ligne


L'IRILL organise une journée de conférences et de discussions sur l'enseignement de technologies des logiciels libres dans l'enseignement supérieur. Notre volonté est de réunir des enseignants et des étudiants de l'enseignement supérieur qui ont eu l'expérience d'enseigner ou d'avoir des cours sur une ou des technologies du monde des logiciels
libres pour une journée de partage d'expériences et de réflexions. Une table ronde suivra pour discuter de la mise en place d'activités coordonnées, et notamment sur un projet inspiré de Summer of Code de Google [5], inséré dans le cursus universitaire.

En effet les dernières années nous avons constaté qu'une multitude d'initiatives ont eu lieu dans différentes universités françaises, mais qui sont restées sans ou avec peu de retour d'expérience. Il serait pourtant particulièrement enrichissant pour les autres membres de la communauté universitaire d'avoir des témoignages de ces expériences.
Ainsi nous pouvons nous en inspirer des bons exemples et essayer d'améliorer les pas si bonnes. En tout cas cela nous permettra de généraliser l'enseignement des technologies libres dans le monde universitaire, car nous sommes convaincus par nos propres expériences et par des témoignages [1, 2, 3] que cela apporte beaucoup aux étudiants, aux enseignants aux établissements et aux communautés du libre.

Plusieurs experts sont invités à venir présenter leurs expériences le jeudi 2 février 2012, dans les locaux de l'IRILL à l'antenne parisienne de l'INRIA, 23, avenue d'Italie, 75013 Paris.

Ce sera également l'occasion de réfléchir sur l'initiative lancée par Albert Cohen reprenant des idées de plusieurs personnes, dont Roberto Di Cosmo, Laurent Séguin (Systematic GTLL) et surtout Olivier Berger (Telecom SudParis) nommée Libre Semester of Code (ou LSoC). Cette initiative est inspiré par le Google Summer of Code et cherche à prolonger cette expérience dans le cadre de la formation universitaire. Pour un peu plus de détails vous pouvez allez voir la présentation du LSoC à la conférence fOSSa à Lyon cette année [4].

Cette journée est préparée par:
Roberto Di Cosmo (Université Paris 7, IRILL),
Albert Cohen (INRIA, IRILL),
Ivaylo Ganchev (LIASD, Université Paris 8).


  1. M. Pedroni, & Co, "Open source projects in programming courses", in Proc. SIGCSE, 2007
  2. D. Carrington and S.-K. Kim. "Teaching software design with open source software." In Proc. FIE, 2003.
  3. R. Hess and P. Paulson. "Linux kernel projects for an undergraduate Operating Systems course." In Proc. ACM Technical Symposium on Computer Science Education (SIGCSE), 2010

Debian bug Squashing Party (Wheezy)

openlogo-nd-100.png Next February, 17 to 19th,  a Bug Squashing Party (BSP) is organized at IRILL offices.

The principle of a BSP is to gather Debian contributors to tackle a maximum of bugs in Debian.
This event is also the opportunity for new potential contributors to meet Debian Developers or Maintainers. Numerous regular contributors will attend to this BSP and will help newcomers to fix their first bugs.
For organization reasons, an inscription on the Debian wiki is mandatory.

“Open Source Innovation Catalyst” track

In the context of the OW2 conference, the "Open Source Innovation Catalyst" track will be organized by IRILL with support from ICT Labs ( . This conference will be held on the november 24 th at Orange labs at Issy Les Moulineaux

Organized by IRILL ( with support from ICT Labs (

Software is a pervasive element in research, not only in Computer Science, for which it is natural to produce software artifact, but in all other disciplines too, where the need to manage and elaborate information pushes researchers to develop and maintain significant software tools.

Open Source software coming from these research activities is a really interesting object for technology transfer, which is essential to produce the innovation on which the competitiveness of modern industry relies. By simplifying matters related to intellectual property, and building collaborative communities, Open Source software lends itself very naturally to fluid transfer and innovation activities. It is thus possible to gain significant productivity in the building of complex software systems, reducing the R&D costs and increasing software reliability.

Good practices in technology transfer using Open Source, as well as difficulties one may encounter in the process, are the key subject of this track.
The programme is the following:
Roberto Di Cosmo, IRILL and Patrick Moreau, Inria
OtaSizzle - an Open Platform to Transfer Technology from a Research Project
to Practice
Olli Pitkänen, Helsinki Institute for Information Technology HIIT
Kassi - creating business opportunities from an open source research project
Juho Makkonen, Kassi
Possible industrial use cases of KALIMUCHO
Philippe Roose, Associate Professors, LIUPPA/Laboratoire d'Informatique de l'UPPA
Polarsys (previously OPEES) as a mean to catalyse the maturation process
(from TRL3 to TRL6) for OSS projects in the Embedded Software space.
Gael Blondelle, OBEO
Open Source Innovation Factory
Paolo Ceravolo, Assistant Professor, Università degli Studi di Milano, Italy, in
collaboration with Engineering Group
Open source and Standardisation
Jamil Chawki, Orange Group
An Open Source-Based Approach for Industrializing Research Tools
Hugo Bruneliere – R&D Engineer – AtlandMod Team (INRIA, EMN & LINA)
Creating Value with Open Source
Teodor Danciu, Founder and Architect JasperReports, Jaspersoft

Sysadmin: episode 3

Sysadmin oriented conference

From October 20th to the 21th, and for the third time, the french sysadmin community will gather to discuss about various topics related to sysadmin.

In a world where free software is the key, IRILL is glad to host this conference. 

Full program (French)

Inscription à l'atelier (EC2 étant utilisé, une participation est demandée)

Inscription aux conférences

Antelink/antepedia presentation by Guillaume Rousseau

On the October 7th at 3PM, Guillaume Rousseau will present the work done by Antelink at IRILL.

Antelink/antepedia presentation by Guillaume Rousseau The management of the whole software development supplier chain of has seen a dramatic increase of complex sourcing that combine internal IT resources from contractors, both onshore and offshore providers, and use of Open Source. In this talk we will present, Antepedia, the biggest Open Source projects knowledge base available, addressing some of technical challenges, we are facing at Antelink setting up Antepedia, and open the discussion regarding key issues such as

  • open source compliance,
  • multi-source tracking from open source components to distro provider like debian community,
  • Software quality assessment related to multi-source tracking,
  • R&D opportunities based on Antepedia.

Speaker : Guillaume Rousseau, Antelink CEO and co-founder. Guillaume is a former assistant professor from University Paris Diderot. He worked 3 years at INRIA Paris Rocquencourt Research Center, working on IP tracking Methodologies and tools. In 2009, he created Antelink (, a spinoff from INRIA

Antepedia ( is the world biggest Open Source projects knowledge base with more than 940.000 open source projects and more than 210.000.000 files that are downloaded from  GoogleCode, SourceForge, Maven Central, Debian repository and other sources such as savannah, RubyForge and OW2.

First LibreOffice conference

The first LibreOffice conference will be hosted by IRILL and by La Cantine. Those who are interested in the development of free office productivity software, open standards, and the OpenDocument format generally should join this event.

 With more than 150 people expected, this first project annual meeting will gather all the LibreOffice community. 

Various tracks (technical, marketing, etc...) will be held during three days, from the 12th of October to the 15th.

IRILL is proud to host this first conference of this major project.

More information:
The official website feed
Twitter feed

IRILL sponsors the video events of Debian

IRILL sponsors the video events of Debian 
One of IRILL goals is to foster collaboration between researchers and Free Software developer communities, and making quality video material readily and publicly available is one important means to achieve this goal.

We have recently acquired professional video material in order to record and broadcast the events we host, and we agreed to lend to the Debian video team this equipment for major events like the yearly Debian Conference (debconf) or FOSDEM in Bruxelles, and to provide storage for the Debian video equipment when it is not used.

We are thrilled to join forces and share resources with Debian, whose video team has an impressive track record of service to the community : stay tuned for  more news about this collaboration over the next months.

Guest Information

Here are some information for our guests.


23, avenue d'Italie,75013 Paris, France

More information

During the week end, some badges will be provided. However, you should create groups to get into the building. During week days, just ask at the reception.


There are two networks at IRILL:

  • the IRILL wifi network: Network name "Irill"  the password is provided on site . The reception might be bad in the Algorithme room.
    Good thing about this network is that there is no filtering
  • the INRIA wifi network: Network name "INRIA-guest". The login and password are provided by the reception.
    This network is filtered


There are many places to get good food around:

  • Pho 14 - 13, Avenue de Choisy. Vietnamese food
  • Mondol Kiri - 159 Avenue de Choisy. Cambodian food (Better book in advance: 01 5379 7596)
  • Via Italia - 21 avenue Italie. Italian food
  • Thaï Yim - 14 Rue Caillaux. Thaï food
  • Le Sukhotai 12 Rue Père Guérin. Thaï food
  • La Taverne des poêtes - 52 avenue des Gobelins. French (britany) food. Crêpe
  • A609. 3, place des Alpes. Nice French Food.



Obviously, on the other side of the road, there is the big mall called Italie 2. A lot shops can be found.

The Tang frere asian supermarket can be found on 168 avenue de choisy.


Some hotels close to IRILL:

You may also find interesting deals at the accomodation site of the Paris Visitors and Convention Service. Choose the district Place d'Italie - Tolbias - Bibliothèque to be in the neighhborhood of IRILL.

IRLL hosts some GCC Compile Farm nodes

IRILL is glad to contribute to the GCC Compile Farm by providing some nodes.

The GCC Compile Farm is an infrastructure which allows any free software developer to get access through SSH to a wide range of machines.

Many architectures under various operating systems are offered.

The annoucement

Description of the GCC Compile Farm


Darcs sprint

April 1st, 2nd and 3rd 2011, IRILL will host the bi-annual meeting of Darcs. Darcs is a distributed, advanced revision control system written in Haskell.

Darcs sprintSeveral topics will be discussed during this meeting. For example, the organisation for the GSoC 2011 or the Seamless Git/etc integration.

More information:
Darcs website
The meeting page


Eigen 2011 meeting

Eigen 2011 meeting  Hosted by IRILL, developers of Eigen, a C++ template library for linear algebra: matrices, vectors, and related algorithms, will gather at IRILL from Friday March 18 to Sunday March 20, 2011.

More information:
The Eigen website
The meeting page


GNU Hackers Meeting

GNU projectFrom August 25th to 28th 2011, IRILL will host the GNU Hackers Meeting

This meeting allows maintainers and active contributors to gather and discuss.

More information: 
GNU Project
GNU Hackers Meetings
GNU Hackers Meeting in Paris


Going to the plenary meeting of the GTLL group in Systematic

<Going to the plenary meeting of the GTLL group in Systematicp>Many of us are joining the plenary meeting of the Free Software Group in Systematic, on February 3rd 2011

Come and meet us , from 16h to 19h at EPITA (Amphi 4),  14-16 rue Voltaire Kremlin Bicêtre


IRILL Days 2010: videos are online!

The IRILL videos for the IRILL days 2010 are now available online, in OGG/Theora, an open format, or H264 as a fallback.

You can see them online here !

GCC meeting at IRILL

<GCC meeting at IRILLp>27th January 2011, some GCC developers will gather to discuss about new features in GCC and MELT but also develop on GCC.

The discussions will be held in French and English.

Contacts: Basile Starynkevitch or Albert Cohen

GCC Homepage

GCC MELT Homepage

Education with and to Free and Open Source Software

Computers & software are becoming pervasive, and are having directly or indirectly a significant impact on almost all aspects of our modern life: it comes as no surprise that computer & software have an impact on education too. Academia has an important role to play at several levels…

  • We need to educate the teachers,
  • We need to teach all university students
  • We need to teach IT students how to work in an environment that is becoming dominated by Free & Open Source software.

At IRILL, we want to contribute  to advance this agenda, by cordinating the actions on the FOSS curriculum.

In this page, you will find links to relevant resources, and to events and actions on this topic.









How to get to IRILL


Avenue d'Italie 23, view from the street IRILL is located in the 13th arrondissement (district) of Paris. The exact address is

23, avenue d'Italie,75013 Paris, France

On site

After entering the building continue down the long corridor on the right side of the hall, and proceed to the escalators on the far side of the building. IRILL's offices are located on the third floor.

Visitors who do not have a badge have to register at the front desk when entering the building.

Acces by Public Transport

The nearest métro (subway) station is Place d'Italie which lies on the métro lines 5, 6 , and 7. Take exit number 2 (Grand Écran) or 3 (Vincent Auriol). Head into Avenue d'Italie which departs south from Place d'Italie; the number 23 lies on the left (east) side of the street after about 200 metres. The entrance is to the left of the post office that occupies the main part of the building front (look out for the sign "INRIA")

In case you are arriving on métro line 7 from the south it may be better to get off at métro station Tolbiac. Take exit number 2 or 3, then walk down Avenue d'Italie in northern direction. The building is on the right side of the street.

Access from the airports

Access from the train stations

Access by Vélib 

The nearest Vélib stations are

IRILL Days 2010

On the 4th and 5th of October, IRILL hosted the IRILL Days 2010, a collaboration event with a rich program, featuring talks from leading researchers in Free and Open Source Software projects like Coccinelle, Ocsigen, Mancoosi and GCC, communities like OW2, and industries working in the FOSS arena.

Videos of the event will be online soon.

Towards advanced education in and with FOSS

Fossa MutualismAt the Education Session of Fossa 2010, several IRILL members will be present and present their experience in teaching using FOSS and about FOSS in computer Science curricula, as well as more general curricula in University. It will also be an occasion to learn about interesting experiences ongoing in various different places in Europe, and in the world.

See you there on November 9th!


25th anniversary of FSF

It is today the  25th anniversary of the Free Software Foundation, and we are all quite happy, here at IRILL, that we start our activities on this very date, with the IRILL Days.

It has been a long way for Free Software, from the very first definition of Free Software licences, then the uptake of the Free Software phylosophy and the acceleration of the development made possible by the generalisation of the Internet, and finally the main inroads made by Free and Open Source software in all areas of IT, even in companies that considered Free Software pure evil just a few years ago.

At IRILL, we believe it is now time to make some extra steps: research on Free Software, especially on how to master code complexity, education to and with Free Software, and setting up ecosystems mixing academia, communities and industries.

This is our mission for the next years.


Enlightening! Free Software research in context

The freedom to use, study, share and improve the programs that we  rely on every day is not just a philosphical ideal; it's also very  good business. Over the past 25 years, Free Software has reached  nearly every corner of the IT world, and now underpins a 50  billion dollar economy.    For research and education, the potential of Free Software is  obvious.

Free Software gives students a chance to fully understand  how computers work. Free Software communities make it possible for  beginners to work side-by-side with highly trained professionals.  Researchers have access to the most advanced developments of their  peers and to a treasure trove of data.    Yet there is still far too little Free Software in education and  research, just like in society in general. Most schools and  universities essentially require their students to use non-free  programs.

Researchers can contribute to a better understanding  of the workings of Free Software communities. Free Software  advocates need more and better figures on Free Software use and  development in order to convince decision makers. By delivering  these things, researchers can create a better basis for their own  work.

Dynamic Web Development with Seaside

Presentation by Stéphane Ducasse

imagine being able to debug and fix a broken web application in the middle of the request or to reuse control flows and web components with the ease of desktop application development. This presentation will give a quick introduction to a Web framework that is different by design. It will demonstrate new patterns of Web development, that let you build highly interactive Web applications quickly, reusably and maintainably.
Seaside integrates with latest web technologies such as jQuery and Comet. Seaside is developed in the new open-source Smalltalk pharo in which the RMOD INRIA team participates actively in. We will present briefly
Pharo which is use as developement environment by Gemstone - now VWMare.

IRILL hosts the OSI Board meeting

IRILL hosts the OSI Board meetingJust after the Open World Forum, the OSI board meeting will take place in Paris, on the IRILL premises. It will be one more occasion to foster collaboration between researches, developers, communities, industries and educators on Free and Open Source Software, which is the mission of IRILL.


IRILL welcomes the FSFE at Irill Days

IRILL welcomes the FSFE at Irill DaysIn the opening session of the IRILL Days, on October 4 and 5, we are honoured to welcome Georg Greve, founder of FSF Europe, and Karsten Gerloff, the current president of FSFE. FSFE has done an important work defending Free Software in Europe, and continues doing so. It has also collaborated with various academic actors in the framework of some european projects.

Debian Release Team meeting at IRILL

Debian Release Team meeting at IRILLIRILL is pursuing a rather ambitious goal: finding and establishing synergies among researchers, industries, and communities who share a common interest in Free Software. In doing so, IRILL does not want to miss chances to support FOSS communities with its resources, in order to give something back to the members of such communities who enthusiastically contribute to Free Software projects. IRILL believes that organizing community meetings is a sensible way to help Free Software and will start doing so since the very beginning of IRILL activities.

We are hereby happy to announce that IRILL will sponsor and host a Debian Release Team Meeting on October 2nd and 3rd, in its INRIA premises in Paris. The Debian Release Team is busy finalizing the next Debian release, codenamed Debian Squeeze. We are grateful for their work and for that of all members of the Debian community. If you are a Debian enthusiast as we are at IRILL, stay tuned on the usual Debian channels for more information about the meeting and on how to help the release of Debian Squeeze.

IRILL Days: where FOSS developers, researchers and communities meet

On October 4th and 5th 2010, IRILL hosts the first edition of IRILL Days, on the Inria premises, at the 5th floor of 23, Avenue d'Italie, Paris.

This event sports a rich program, with presentations by researchers working on FOSS, developers and industries, as well as community representatives, and quite a few networking opportunities on IRILL topics.

It will be the first public event of IRILL, and it's open to everybody. We just ask you to register in advance, due to limited space availability.


Irill presentation at the Open Innovation session in OWF 2010

Keynote at OWF 2010

A presentation of the philosophy behind IRILL, and some of the first concrete actions done in the framework of this initiative will be given in a keynote of the Open World Forum in Paris, on September 30th 2010.



IRILL website credits

This website brought to you thanks to:

IRILL - Research and Innovation on Free Software

Initiative de Recherche et Innovation sur le Logiciel Libre



Free  and  Open  Source  Software  provides  the  fundamental  building  blocks  for  most of today's and tomorrow's software solutions, being them commercial or not. FOSS has become ubiquitous.

FOSS  software  components  are  inter-connected.   While  writing,  compiling,  testing,  installing  or  customizing a component, a developer is  often  led  to  propose  modifications to other  FOSS  components that are outside of his direct reach. Software has become transparent: complex systems built out of FOSS components can be opened up and  fully  scrutinized,  from  the  system  level  down  to  the  single  memory  word of data  or  code, with no artificial barriers imposed by the absence of access to the sources of vital components.

FOSS has changed for ever the way how software is developed, maintained, and deployed. Affordable high-speed internet for everyone, and the applications coming with it, have paved the way for new interaction patterns between users, developers, researches and other stakeholders.  FOSS developers communicate and collaborate via any combination of e-mail, instant messaging, phone and video-conferencing. They share information either directly or through wikis, forums, or indirectly, through search engines or news aggregators. Collaboration between geographically distributed developers, of paramount importance for FOSS, goes hand in hand with new collaborative tools like distributed version control systems, or software forges. New degrees of collaboration between developers need new generations of tools.

Software development as FOSS raises questions that go beyond the problems known from proprietary software: some new issues are raised by the different development process; others are due to different economic models, and some very challenging ones originate in the new possibilities opened by the fact that software is now becoming transparent.


The mission of IRILL, the Center  for  Research and  Innovation  on  Free  Software, is to bring together in one place leading  researchers  and  scientists,   expert   FOSS  developers,  and  FOSS  industry  players  to  tackle  the  three fundamental challenges that FOSS poses today:

  • scientific: study, explore and solve the new problems raised by the development, maintenance and wildly varying evolution process of the large mass of code that FOSS gives access to
  • educational: adapt curricula for users, system administrators, and developers to prepare them for a computing infrastructure in which FOSS plays a prime role.
  • ecomomic: contribute to create a sustainable ecosystem for the FOSS innovations

IRILL's mission is to foster exchange of knowledge between FOSS developer communities and computer science researchers: new problems coming from the FOSS world will feed the work of researchers who are eager not only to find innovative solutions, but also to turn these solutions into tools that will improve the real-world, daily work of the FOSS communities. Developers and researchers will bring in their joint expertise to create new course materials for computing courses.

Keeping in touch

You can receive news about IRILL events by subscribing to the irill-news mailing list, subscribing to the RSS feed and following us on Twitter,, Facebook, LinkedIn and Google plus.


Irill is created by:

University Pierre et Marie CurieLogo Inria rougeLogo Paris Diderot

with support from:

Logo Ile de France

Going to the plenary meeting of the GTLL group in Systematic

Going to the plenary meeting of the GTLL group in SystematicMany of us are going to participate to the GTLL plenary meeting, on September 15th, hosted by University of Paris 6.

There will be also a short presentation of IRILL. One can register here.

Open Data and the Data Publica Project

Presentation by François Bancilhon

Open Data : the idea of Open Data has been around for quite a while, but it acquire real momentum with the launch of the program by the Obama administration and the program under the inspiration of Tim Berners Lee. In France the CADA law and the set up of APIE have been the initial steps. Other initiatives such as Fing, Regards Citoyens, GFII and Libertic are worth mentioning. The basic idea is that public data, held by public organizations should be available to citizens and economic actors, unless they violate some authorship right, the privacy of individuals or national security. Motivation for this open publication range from the right of citizen to information, the transparency and accountability of government and the positive economic impact.


This will be a brief survey of the Open Data phenomenon from various points of views: historical, political, technological, legal and economic. It will briefly present the key players, the various points of views and the various drivers. The talk will cover the worldwide situation, with a specific focus on the French territory.


Data Publica: The Data Publica project is being developed by a consortium of three startup companies: Araok, an e-commerce company, Nexedi, the editor of the open source ERP software ERP5, and Talend, the world leader of open source ETL. It is partially funded by the French “Secrétariat d'Etat chargé de la Prospective et du Développement de l'Economie numérique”. The objective of the project is to build and manage a market place where public and private organizations which own data can publish their meta-data, the way to access the corresponding data and the license under which it is usable, and Internet and mobile application developers can search and visualize the meta data, create new applications and deploy them by using the data according the licenses granted


This will be a brief survey of the project with data about the current results and a demo of the platform.

Free Software and the Playing Field

Presenton by Georg Greve

In only a few years, Free Software has evolved from being a niche phenomenon into an increasingly mature mainstream movement. Despite the commonplace understanding as Free Software as one of the driving forces of tomorrow's information technologies, the surrounding political and economic environment has often not yet kept up. As the founder and first president of the Free Software Foundation Europe, as well as CEO of a Free Software enterprise, the speaker has unique insight into political and economic aspects that keep favouring proprietary technologies until the current day despite the assurances to the contrary by some. From his personal experience, Georg Greve will give some real life examples of how Free Software companies work and interact with partners and customers, and how a truly level playing field would be constructed.

The IRILL Newsletter is now available

You want to be informed of what is going on at IRILL? You can of course add our RSS feed to your favourite reader, but you can also subscribe to the IRILL Newsletter, which is now up and running!

IRILL website now available

IRILL website now availableThis summer has been a very active one for IRILL. Under the hood, preparation work for the forthcoming IRILL launch has been going on relentlessly. As the first visible milestone of such work, we are hereby proud to announce that the IRILL website is now available at

The website is still undergoing heavy editing, but it already contains important material such as the first projects that IRILL will host, information about IRILL founding members and partners, forthcoming events, etc.

Have a look and let us know what do you think!

IRILL Days 2010: detailed program


Irill Days 2010

4,5 October 2010,

23 Avenue d'Italie - 5th floor

Paris, France 



Videos of the various talks are available.

Monday 4th October 2010


9h-9h30 Registration and breakfast

9h30-9h45 Welcome address


9h45-12h15 General session: Free and Open Source Software: challenges for research, economy, and education


9h45-10h15 Michael Tiemann, President, OSI and VP RedHat

10h15-10h30 Venkatesh Hariharan, RedHat

10h30-10h45 Coffee Break

10h45-11h15 Karsten Gerloff, President, FSFE Enlightening! Free Software research in context

11h15-11h45 Georg Greve, Kolab Systems, CEO; FSFE, Board Member: Free Software ant the Playing Field

11h45-12h15 Roberto Di Cosmo, Director, IRILL


12h15-14h00 Lunch break


14h00-16h00 Technical session: Early projects at IRILL

14h00-14h30 Mancoosi: managing your GNU/Linux installation, Stefano Zacchiroli

14h30-15h00 Coccinelle at the service of the Linux Kernel, Julia Lawall

15h00-15h30 Ocsigen: rethinking web programming , Vincent Balat

15h30-16h00 Coffee Break


16h00-17h00 Demo and speed dating session


End of day one


Tuesday 5th October 2010


9h-9h30 Registration and breakfast


9h30-12h15 Technical session: Emerging projects and technologies

 9h30-10h New Trends in Scilab, Claude Gomez

10h-10h30 Building an R&D ecosystem around GCC: what, why and how?, Albert Cohen

10h30-10h45 Coffee Break

10h45-11h25 Component-based programming with Fractal - models, tools and the FraSCAti platform , Philippe Merle

11h25-11h55 Dynamic web development, Stéphane Ducasse


11h55-12h20 Demo and speed dating session

12h20-14h00 Lunch break


14h00-16h30 Technical session: Industry and innovation

14h00-14h30 Hi-Lite: a Verification Toolkit for Unit Test and Unit Proof, Yannick Moy (Adacore)

14h30-15h00 Open Source Cloudware initative: towards a free, open cloud, Jean-Pierre Laisné (OW2)

15h00-15h30 Open data and the Data Publica Project, François Bancilhon

15h30-16h00 Coffee break and speed dating

16h00-16h30 DORM, the missing tool to persist your deliveries, Pierre Queinnec (Zenika)

16h30-17h00 A short survey of Real Time Linux technologies, Pierre Ficheux (OpenWide)


End of Irill Days.



IRILL contact information

For all information and inquiries about IRILL, please mail us at:

IRILL is located at 23, avenue d'Italie,75013 Paris, France, check the access page for more information.

New trends in Scilab

Presentation by Claude Gomez

Scilab is the Free Software for Numerical Computation. Born of INRIA research, it is developed now in the frame of the Digiteo Foundation and Scilab is ready for both Academics and Industrial use. But new important  evolution is arising. First, a major evolution for Scilab software with new release Scilab 6.0 coming in 2011: Scilab 6.0 will have a brand new  kernel. Second, another major evolution for Scilab organization with the creation of a company, Scilab Enterprises, which will take care of all  the Scilab operation in the future. In this presentation, we will explain why all these changes are essential for Scilab future.


Kill bugs before they hatch.

Coccinelle is a program matching and transformation engine which provides the language SmPL (Semantic Patch Language) for specifying desired matches and transformations in C code. Coccinelle was initially targeted towards performing collateral evolutions in Linux. Such evolutions comprise the changes that are needed in client code in response to evolutions in library APIs, and may include modifications such as renaming a function, adding a function argument whose value is somehow context-dependent, and reorganizing a data structure. Beyond collateral evolutions, Coccinelle is successfully used (by us and others) for finding and fixing bugs in systems code.


The development of Coccinelle is supported in part by the Agence Nationale de la Recherche (France) under the contract ANR-09-BLAN-0158-01 and the Danish Research Council for Technology and Production Sciences.

The development of Coccinelle has been supported in the past by the Agence Nationale de la Recherche (France) under the contract ANR-05-BLAN-0383, the Danish Research Council for Technology and Production Sciences, and INRIA via the cooperative research initiative ARC VeriTLA+.

IRILL Days 2010 - registration

Participation to IRILL Days 2010 is now closed, but you will be able to find the videos of the event online soon.

Component-based programming with Fractal - models, tools and the FraSCAti platform

Presentation by Philippe Merle and/or Lionel Seinturier

A presentation of the issues related to component based programming, in the framework of the Open Source FraSCAti platform.

Ocsigen: Rethinking Web Programming

Presentation by Vincent Balat

The advent of the Web 2.0 and HTML5 changed radically the behaviour of Web sites. Most Web programming tools are not fully adapted to this evolution. The Ocsigen project aims at creating a new integrated way to design modern Web sites as distributed applications. It introduces an abstraction from the underlying technologies in order to make very easy to program complex Web interaction. It also makes use of advanced static typing features to ensure the robustness of programs and the compliance with standards. The talk will give an overview of the project and a demo of how to build a small application.

Mancoosi project highlights: managing your GNU/Linux installation

Presentation by Stefano Zacchiroli

Free and Open Source Software distributions raise difficult problems both for distribution editors and system administrators. Distributions evolve rapidly by integrating new versions of software packages that are independently developed. System upgrades may proceed on different paths depending on the current state of a system and the available software packages, and system administrators are faced with choices of upgrade paths, and possibly with failing upgrades.

While the predecessor project EDOS had focused on tools for the distribution editor, the Mancoosi project aims at developing tools for the system administrator. We pursue two main avenues:

  1. Develop mechanisms that provide for rollbacks of failed upgrade attempts, allowing the system administrator to revert the system to the state before the upgrade

  2. Develop better algorithms and tools to plan upgrade paths based on various information sources about software packages and on optimization criteria

DORM, the missing tool to persist your deliveries

Presentation by Pierre Queinnec

The software configuration management (SCM) systems are used to share  source code. During a build process, source code produces objects,  derived objects. These objects are often temporary; however, it's not  always the case; for example framework libraries. Additionally, these  frameworks need to be shared across all the teams. Therefore, there is a  need for a persistence layer, implemented through repositories and  management metadata. Today, both the FOSS community and the proprietary  world provide only repositories for objects produced for the Java  platform. Consequently, we plan to provide a Derived Object Repository  Manager (DORM) managing all kinds of objects.

The presentation will discuss the following key questions:

  • Why do you need a repository manager?
  • What is the novel approach for derived objects' persistence within DORM?
  • How do you use DORM and extend it for your needs?
  • What is the DORM roadmap?

A short survey of Real Time Linux technologies

Presentation by Pierre Ficheux

After shortly recalling the basic concepts of hard real time systems, we show why a standard Linux kernel is not well suited for these kinds of applications. We will then present the main free software solutions for hard real time on Linux systems: PREEMPT-RT, RTAI, and Xenomai. We will also give a short demonstration on an Intel/Atom platform.

Coccinelle at the service of the Linux Kernel

Presentation by Julia Lawall

Coccinelle is a tool for program matching and transformation in C code,  targetting the needs of Linux developers.  A key feature of Coccinelle is  a specification language that has been designed to be close to the C  programming language and the patch syntax.  Unlike the case of standard  patches, however, the Coccinelle matching process takes into account the  semantics of the Linux code, and thus Coccinelle specifications are  referred to as semantic patches.

Coccinelle has primarily been used for  automatic bug finding and source code evolution.  Over 700 patches  developed with the help of Coccinelle have been accepted into the Linux  kernel, and Coccinelle semantic patches are now available to Linux  developers as part of the Linux source tree.

In this talk, we will present Coccinelle and give some representative examples of how it has been applied to Linux.

We will then present some directions for future research on tools to further improve the Linux development process that could be carried out in the context of the IRILL.   Coccinelle is currently being developed at INRIA, supported by the ANR Blanc project ABL, and at the University of Copenhagen, supported by the Danish Research Council project ISIS.

Hi-Lite: a Verification Toolkit for Unit Test and Unit Proof

Presentation by Yannick Moy

Until now, verification of safety-critical embedded programs has been relying mostly on tests. Costly code coverage objectives have been designed for these tests to achieve a reasonable level of confidence. The upcoming avionic standard DO-178C will allow one to replace tests with proofs, thus reaching higher levels of confidence at a lesser cost.
In order to benefit from this updated standard, aeronautic industries will need new development environments which include tools for proving program properties, which have been developed mostly in an academic setting until now. The goal of project Hi-Lite is to integrate existing verification tools (Verification Condition generator, automatic prover, static analyzer) inside industrial tool-chains (GUI, compiler, debugger) for Ada and C. Some of these tools are being developed by AdaCore and Altran Praxis companies, while others are developed by research labs at INRIA and CEA. All these tools are free software.
Free open-source software facilitates collaborations between teams that are expert on their own tools, in particular when academic research labs are involved. This mode of development partially fulfills a demand from industries which are anxious about finding expert support for long-lived projects. We also adopted a completely open development process for the new tools in Hi-Lite, with a public source forge, a public web-site for intermediate documents and a public mailing-list for discussing technical issues, which are all available from the Open-DO web-site.

OW2 Open Source Cloudware initative: towards a free, open cloud

Presentation by Jean-Pierre Laisné

The OW2 consortium decided, in may 2010, to start an initiative to collaborative
develop and maintain open source components, based on open standards, for cloud

More than 30 members worldwide joined already this intitiative, from Europe,
China, Brasin and the USA, while promising connection are being established
with existing projects like Open Nebula, or OpenStack.

We will shortly present the design principles, and the first software components
issued from this intitiative, and we will present the main challenges and the
next steps in this development, looking forward to fruitful discussion and
collaboration with the IRILL members.

Building an R&D ecosystem around GCC: what, why and how?

Presentation by Albert Cohen

What. The GNU Compiler Collection (GCC) is ubiquitous in free software development and beyond. It faces multiple challenges, as a developer community, and on the software infrastructure side.                                                                                                                                                                                        
Why. Parallelism and hardware specialization push compilation technology towards its limits, while dynamic, productivity languages reshape the role of compilers. Both open exciting research avenues, but compiler construction requires rare and skilled developers, while compiler development is generally seen as a necessary cost, but not an  instrinsic source of value                                                                                                                                                                                                                 
How. We believe it is possible to build a more coherent R&D ecosystem around GCC, maximizing the contribution to the platform itself and the impact on programmers (GCC users). Innovation in such an ecosystem takes the form of a combination of mutual-interest development, seed funding,  innovative business models and ambitious technical goals.                                                                                                                                                                                  
We review some directions towards the emergence of a rich, academic and  industrial ecosystem bringing together continuous funding and expert  developers, leveraging recent experience and results. 

Keynote on IRILL at the Open World Forum

Keynote on IRILL at the Open World ForumA presentation of the philosophy behind IRILL, and some of the first concrete actions done in the framework of this initiative will be given in a keynote of the Open World Forum in Paris, on September 30th 2010.

Coming soon: IRILL

The Ice Tube Clock kit

Free Software is a disruptive phaenomenon: free redistribution of the software, free access and modification of the source code, and collaborative developement are in total opposition to the highly restrictive practices that have dominated the software industry for decades.

Free Software is designed, developed, maintained, distributed, marketed, and used in ways that are radically different from those that were customary just a few years ago.

Free Software poses new challenges that must be addressed to allow Free Software communities, developers, industries, and users to collaborate effectively on code bases whose size, variety and complexity are mind-boggling.

A first class of challenges are scientific. The Free Software code base grows at an astonishing pace, and managing its complexity is no longer possible without new research breakthroughs and the developement of advanced tools; as an example, one can consider the number of software packages in Debian, a widely used GNU/Linux distribution made by volunteers, which skyrocketed from a few hundreds in 1994 to 30.000 in 2010; or the number of Eclipse plugins, that is counted in the thousands today.

A second class of challenges is posed by education. As brilliantly stated by D. Patterson we can no longer be contented with engineers that are trained by teaching them only to write code from scratch on their own; we need to develop a new curriculum, fully leveraging the new possibilities that Free Software offers, and this will require a significant effort.

A third class of challenges is to adapt the traditional technology transfer process to the Free Software ecosystem, where strategies based on per copy licence fees are no longer pertinent.


IRILL is a research and innovation initiative that will contribute to address all these challenges.

IRILL is coming soon!


stay tuned on this blog ...


Fresh air in Web programming

The Ocsigen project is aimed at proposing clean and safe tools for developing and running Web 2.0 applications.
It contains a powerful Web server, a full framework for developing dynamic sites, and many other extensions.

Main features

  • An extensible Web server
  • A full set of Extensions for serving static pages, CGI pages, for HTTP redirections, data compression, reverse proxy, access control, etc.
  • Eliom Web programming framework, with:
    • Integrated client and server side programming with automatic bidirectional communication
    • Dynamic creation of services
    • Static checking of XHTML (using OCaml or OCamlDuce)
    • Automatic management of sessions
    • Concise and modular programming
  • Js_of_ocaml: a compiler from OCaml to Javascript to write client applications in OCaml!
  • O'Browser: an ocaml virtual machine in Javascript, alternative way to run your OCaml programs in browsers
  • Lwt: a cooperative threads library for OCaml
  • Ocsimore: a set of libraries to write your Web sites, containing user management, wiki, forum, blogs ... (under development)
  • Macaque: a library for safe and flexible database queries using comprehensions
  • O'Closure: a binding for the Google closure library, to put sophisticated widgets in your applications.



Manage the Complexity of the Open Source Infrastructure.

Mancoosi is a European research project in the 7th Research Framework Programme (FP7) of the European Commission. The project is situated in the FP7 theme Information & Communication Technologies (ICT), Challenge 1: Pervasive and Trusted Network and Service Infrastructures, Objective 2007.1.2 Service and Software Architectures, Infrastructures and Engineering.

Free and Open Source Software distributions raise difficult problems both for distribution editors and system administrators. Distributions evolve rapidly by integrating new versions of software packages that are independently developed. System upgrades may proceed on different paths depending on the current state of a system and the available software packages, and system administrators are faced with choices of upgrade paths, and possibly with failing upgrades.

While the predecessor project EDOS had focused on tools for the distribution editor, the Mancoosi project aims at developing tools for the system administrator. We pursue two main avenues:

  1. Develop mechanisms that provide for rollbacks of failed upgrade attempts, allowing the system administrator to revert the system to the state before the upgrade (work package 2 and work package 3)

  2. Develop better algorithms and tools to plan upgrade paths based on various information sources about software packages and on optimization criteria (workpackages work package 4 and work package 5).

The consortium as a whole is entirely committed to the free software movement. All the software deliverables as well as the intermediate revisions will be freely accessible to the public. Software developed in the context of the project will be made available under a free licence. Software, reports and articles produced in the context of the Mancoosi project will be available through this web site.


Who we are

The goal of IRILL is to take over some of the significant challenges coming from the generalisation of Free and Open Source Software; for this, we want to attract some very special people to join it, and this page tries to give a rough idea of the identi-kit of an IRILL member.


A typical senior researcher at IRILL is a well established and recognised researcher, that has also participated, or followed actively the development of some free or open source software project, and that is ready now to bridge the gap between these two activities. Stepping back for a while from the code and software development, he identifies some significant issues in the evolution of that development, makes them the object of his research, and once some significant improvements come out, will make an effort to bring them to the attention of the developer community of the original project, ideally up to the point of having them adopted. This same criterion applies to whole projects or research teams joining IRILL.

In parallel with the research activity, we encourage activities oriented towards the development of new components of a Computer Science curriculum, specifically taking into account the availability, for  use, contribution and study, of free and open source software.

Researchers, teachers, projects and team are hosted in the center for a limited amount of time.

Community developers

One of the center's foremost activity is the invitation of top notch free and open source developers to come and spend a limited time (typically ranging from two weeks to two months) in the center, to continue their regular activity in the center's rich environment, while interacting with the center's members, via a few seminars, or master classes. From this interaction, research and collaboration ideas will emerge, as well as significant contributions to the teaching materials.

Industry partners

Specific research and development projects carried on by industrial partners may be hosted for a short period in the center, with the same goal of fostering collaboration and innovation, transferring technology from research to industry, and seeding research  with issues from the real world.


IRILL has been created by the following three French institutions

University Pierre et Marie Curie INRIA University Paris Diderot

with support from

Logo Ile de France


Members of the Scientific Advisory Board

The Scientific Advisory Board evaluates the operation of the IRILL, gives suggestions on the lines of research and validates the choices of the projetcs and teams hosted in the Center.

The current members of the Scientific Advisory Board are :

  • Alex Aitken
  • Chris di Bona
  • Manuel Hermenegildo
  • Jean-Marie Lapeyre
  • Andreas Zeller


How we take decisions

The IRILL governance is made up of the following bodies and roles:

  • Board of Trustees (Collège des membres fondateurs): composed of representatives of the founding members, it oversees the operation of the IRILL, validates the members of the SAB and the SC, and appoints the Director
  • Scientific Advisory Board (Conseil Scientifique): evaluates the operation of the IRILL, gives suggestions on the lines of research and validates the choices of the projetcs and teams hosted in the Center
  • Director: heads the IRILL, taking all the decision necessay to ensure the mission of the IRILL is accomplished

In addition to these bodies, the IRILL has a Board of Partners (Collège des membres associés), open to the industries that want to be in tight contact with the work done at IRILL.

IRILL founding members

IRILL founding members

IRILL has been created by the following three French institutions

University Pierre et Marie Curie Logo Paris Diderot Logo Inria rouge


Supported by ...

Logo IDF

Announcement of the creation of IRILL at OWF 2009

INRIA press release on Octobre 1st 2009

Open World Forum logoParis, October 1st 2009

During the Open World Forum, INRIA clearly showed its continued support for FOSS, and announced the creation of the CIRILL. Read more about this in the official press release.

Document Actions