Mandrakesoft to lead new research project in software development
Tuesday December 21 2004
Paris, France; December 21, 2004.
Major European research institutions and Open Source software companies today announced the launch of EDOS, a project dealing with complexity management in the field of Open Source software. The participants will collaborate in the development of theoretical and technical solutions to the management of large-scale, modular software projects. EDOS will receive EUR 2.2 million in European Union funding, in a total budget of EUR 3.4 million.
Software projects have grown to unforeseen levels of complexity. For example, most recent Linux operating systems are comprised of thousands of individual packages. This makes putting together such a system a difficult task; and the short release cycles traditionally practiced in Open Source software mean that this task has to be constantly repeated. This is the case, not only for Linux, but for any large, modular software project.
Consortium members are more specifically concerned with the issue as it appears in Open Source projects. As the amount of Open Source software grows, so does the problem of complexity. That does not mean, however, that nothing can be done about it. Today the building and testing processes are only partly automated, leaving much room for improvement in productivity and quality testing. Participants in EDOS, which stands for Environment for the Development and Distribution of Free Software, will research and implement new ways to manage complexity in Open Source software. The team includes 6 research and academic institutions and 4 software and services companies (see full list below).
Two particular applications are planned. The first one deals with the build process: a distributed, peer-to-peer application, designed to aid the integration of the many software parts that make up a Linux system, will be developed. It is expected that the move from a centralized build and storage process to a distributed one will greatly increase efficiency, because storage and computing capabilities will increase exponentially, and so will network throughput. Resources are not the only problem at hand: intricate problems of dependencies between units inevitably arise in large projects, with every part depending on many others for their proper operation. EDOS members intend to find ways to deal with this problem in a systematic way, using the tools of theoretical computer science.
"There's an issue any large project has to deal with at some point, and that's how to manage complex dependencies between parts so as to get an integrated, coherent whole.", said Roberto Di Cosmo of Paris 7. "This is getting crucial for Open Source software projects as well, because more and more time is being wasted on management issues. We will propose a solution to this problem using formal methods from theoretical computer science, a research area which is particularly strong in France".
The second one deals with quality testing. Free software today lacks a test suite that is both automatic and comprehensive. Testing a Linux OS, or indeed any large application built on Free/Open Source software, is a time-consuming and essential operation. Part of EDOS' plan is to develop tools to make testing more efficient and more comprehensive. A unified, computer language-agnostic framework will allow developers to easily create test modules for their software. Overall quality should greatly improve as a result.
"It is next-to-impossible to manually test every single function in a large piece of software", declared Anton Anisimov of SOT. "Having a program do it is only natural. A comprehensive, effective test framework will save developers, testers and users a lot of time and will save organizations a lot of money. We believe this will be a major contribution to future software development".
EDOS will receive European Commission funding in the amount of EUR 2.2 million. The project runs for a period of 30 months. The resulting software will be released under an Open Source license, and theoretical results will be published in scientific journals as advances are made.
"The concrete results EDOS will bring about will benefit everyone, not just Open Source developers. EDOS brings together some of the field's most prominent players, to solve the problem of complex dependencies before it becomes unmanageable and to improve testing processes. This will result, in the short term, in better software. Ultimately, it will also contribute to Europe's competitive edge in the world market", concluded Francois Bancilhon, Mandrakesoft CEO.
Research and Academic Institutions
- INRIA, France (Project Coordinator)
- University of Paris 7, France
- Tel-Aviv University, Israel
- University of Geneva, Switzerland