SCOS at WorldComp’08
Consortium presented its Strategic and Economic Study of Scientific Computing at the International Conference on Scientific Computing (CSC’08), July 17th in Las Vegas.
LAS VEGAS - 17 July 2008
SCOS presented the results of its “Strategic and Economic Study of Scientific Computing Software” and the paper to appear in the International Conference on Scientific Computing (CSC'08).
The main results of this SCOS study on 212 scientific computing software are:
1) Mature source codes: The vast majority of the source codes studied received the highest rating on the “Maturity” criterion with an average of 5 points out of 6. Implicitly, the same holds for their stability. Finally, no code has had a tumultuous history, or has crossed a significant crisis. That in itself is a sign of quality.
2) Communities and Open Source: Information from users are often unavailable if they actually exist. Few discussion forums have been created to foster a community spirit. When they exist, they are rarely referenced and therefore little used, even if some codes are listed on the most popular search engines (freshmeat or Source Forge, for example). Google, one of the most popular search engine on the Internet, is often of little help due to the low referencing of tools related to scientific computing. Consequently, finding a known code becomes a difficult task that does not helping promoting the usage of the software looked for. Most open source software mentions this issue and complains of a too small community to ensure their survival over time. A better communication and developing a “community spirit” (for developers and users) should therefore constitute a priority to work upon.
3) Services: French source codes exhibit a significant delay in terms of services (training, advise, support) compared their proprietary and international counterparts. The lack of facilities often prevents researchers to meet users’ expectations. But on the positive side, French codes are well documented and kept up-to-date. This represents a strength that should be promoted.
4) An ongoing activity but often hidden: Most software activity is related to their development. A code evolves, mutes, improves and becomes more efficient or falls into oblivion. The history of software is kept track of in its ”releases”. They are often referred to but little information is available on which develops what and especially when. Similarly, the road maps are rarely published. When they are available, they are rarely dated, although this information is important in a community management. To conclude, projects management often lacks clarity and readability over time.
“Only 103 codes could be evaluated in an initial base of 212, due to dead links or inexistent Web site” reveals Jean-Marie Davesnes (Oxalya, France), in charge of the study. He explains: “This figures show the very huge disparity between some really strong codes and a large number of un-structured niche codes”. Sonia Waharte (University of Waterloo, Canada) who co-authored the paper, performed the presentation at CSC’08.
For Alban Schmutz, Coordinator of the SCOS project ”The really great job done in that study emphasis the work that have to be pursued by the scientific computation community: make the code known and give them a solid structure to answer the scientific needs!”. “The SCOS project will bring its forces to build a wide cross-domain and cross-sector community to address that challenge in the following years”.
About SCOS/OSCOS
Scientific Computing has to enter the world of industrialized solutions. Shared standards and interoperability should be the rule, no the exception any. This is the aim of the OSCOS/SCOS project: making scientist's life easiest, sharing tools and results with its colleague when needed.
To achieve this aim, is structured SCOS into 4 WorkGroups: System Interoperability, Applicative Interoperability, Industrialization Tools, Support Team.
SCOS/OSCOS involves 23 industrial and academic partners invested at an international level to support scientific computing software strengthening. The project is co-funded by the French National Research Agency (ANR) and supported by the SYSTEM@TIC-Paris-Region competitiveness cluster.
More information: oscos.org




