Chaired by: Gerrit, scribed by Dave
- Dave Raggett, W3C
- Fabio Paterno
- Carmen Santoro, ISTI
- Davide Spano, ISTI
- Gerrit Meixner, DFKI
- Paolo Bottini, University of Rome
- Gaelle Calvary, LIG
- Joelle Coutaz, LIG
- Cristina Cachon, CTIC
- Jaroslav Pullman, Fraunhofer
- Publication of the task model document
- Progress on the AUI document, additional editors?
- Progress on Introduction document
Publication of the task model document
This is the most important topic for today.
Fabio: thanks everyone for their comments.
One question is the name of the specification: we agree that the starting point is CTT, and of course there have been some changes. We see two options: a) to call this CTT and b) to call it the W3C task model specification. From ISTI point of view we don't have a strong opinion. If (b) we would continue to update CTT independently.
In regards to the names of editors, ISTI have done most of the work, even though others have contributed ideas, and hence are listed in the acknowledgements.
Rather than have an on going academic discussion, we need to to focus on getting something published soon.
Dave: this would be just a first public working draft, and we can make changes in subsequent versions. We need to publish something soon as we are late according to our charter.
Fabio: let's address Gaelle's comments first.
Regarding the synchronization concept, we can improve the description.
Regarding the operators, it is true that they all associative, some are commutative, some are not. We should it this clear which operators are commutative in the specification.
Discussion on how to deal with the context of use. Fabio would like to focus on standardizing the more stable aspects rather than areas subject to on going academic study.
Gaelle: suggest removing the ??? attribute.
Fabio: we could put some of the more speculative work into a separate document.
Paolo: agree with separating context of use into a separate document, but wouldn't agree with treating this with operators. Also question of optionality (zero|one vs one..N). My email today suggests having two operators. See figure attached to following email:
Fabio: we need to change the enumeration of subtasks, e.g. to add "0, 2..*" where appropriate, where currently we have "2..*". A single subtask is not permitted.
Question about the condition group, we can try to make it clearer.
Gaelle and Joelle: we agree with what needs to be expressed but don't think the current UML diagram does this correctly.
Some clarification discussion ...
Gaelle: you should include logical unary operators
Fabio: okay we can add those.
Joelle asks whether we really need the abstraction category for tasks.
There could be open enumeration of task categories.
Fabio: we could extend the abstraction category with a subclass property.
Discussion switches to the name of the specification. Gaelle would like to clarify this.
Dave would like to avoid there being two very similar languages, better to agree that the W3C Recommendation defines a snapshot of CTT, and we would aim to update the Recommendation to track changes to CTT as these stabilize.
Gerrit: we need two implementations to become a Recommendation.
Paolo: this specification defines an abstract notion whereas CTT provides a concrete visual notation.
Fabio: the abstract metamodel would be expected to evolve as well.
Dave: we could do a straw poll, but we first need to clarify the implications of the choices.
Fabio: we could follow Dave's suggestion that this updates CTT and make it clear what the relationship is with former definitions of CTT.
Fabio turns to the comments by Gerrit and Paolo re suspend and resume, where one task can interrupt another.
Gaelle: distinction between interleaving and suspend/resume. The specification needs to make the semantics clearer.
Fabio: regarding the user tasks, these are valuable for usability analysis
Gerrit has proposed some clearer wording for task types, this is helpful.
Gerrit proposes invariants in addition to pre and post conditions.
Fabio: we are not specifying a programming language, and don't feel that invariants are appropriate at the level of CTT.
In regards to one of Paolo's comments, Fabio says that with the UML meta model, this is actually quite a precise specification.
Paolo: it is mostly a matter of making the semantics clearer. However, we would like a clear way to map other task modelling languages into CTT.
Fabio: we will integrate these comments and want to go for a decision to publish in next week's call.
Dave to follow up with Fabio on editing the html spec rather than the Google doc.
Gaelle/Joelle: some problems with mailing list, asks Dave for help and progress on the Invited Expert application status.
Dave to follow up separately.
We run out of time, and the meeting is closed