|
Vendors present:
- AmsterCHEM represented by Jasper van Baten (part time)
- HTRI represented by Joe Holmes
- Infochem Computer Services represented by Richard Szczepanski
- Netral represented by Thomas Capricelli
- PSE represented by Thomas Williams
Management Board members present:
- BASF represented by Ronald-Alexander Klein, Werner Drewitz, Frank Güttner (part time)
- Shell represented by Ray Dickinson
- Total represented by Francis Luck
Invitees:
- MPT represented by Anaïs Cassajus
- Peter Banks Associates represented by Peter Banks
Ray Dickinson introduced the session with some background slides that posed the following questions:
- Is CO-LaN interoperability testing useful?
- Should CO-LaN consider CAPE-OPEN compliance endorsement and what would this mean in practice?
- What would help to speed up the process of achieving CAPE-OPEN compliance?
1. CO-LaN Interoperability Testing
-
Vendors tend to concentrate their testing on combinations involving the major simulators, so CO-LaN testing of less common combinations is useful and does identify
worthwhile issues. Also some vendors do not have access to a range of PMEs.
-
Vendors could be proactive in providing CO-LaN with specific test sets that exercise features that they know are important for their products.
-
CO-LaN should consider whether an automated testing system could be produced to enable tests to be repeated easily when updates are produced and also to address the
combinatorial problem that is already arising.
2. Compliance endorsement
-
There was little enthusiasm for formal compliance endorsement on the grounds that, at this stage, it would be difficult to guarantee PMC A would actually work with PME B,
even if both were endorsed as compliant.
-
There was some support for the publication of the CO-LaN test results, although it was clear that there could be difficulties if negative results were published. CO-LaN
does not want to jeopardize the free testing licences that enable it to carry out its interoperability testing programme.
-
A possible solution would be to publish only the positive test combinations, but more consideration is needed into this area.
3. Faster compliance
- Example code was the most important aid to successful implementation.
- A reference implementation, such as COCO, was regarded as more useful than the CO Testers.
- The ability to supply a list of all known problems with a particular software product would help.
- Working interfaces supplied by one or more of the main PME vendors was important, though not in CO-LaN’s control.
This session was felt to be useful, but the timing should be considered for future meetings. The current timing was chosen to avoid clashes with the Workshop and the SIG meetings, but the result was that several vendors did not arrive until after the meeting had finished.
|
