opencsa-liaison — archive

[Sanjay Patil]

1. Roll Call Simon Nash, Bryan Aupperle, Martin Chapman, Anish Karmarkar, Sanjay Patil, Mike Edwards, Jeff Mischkinsky 2. Scribe Assignment Sanjay 3. Approval of minutes of 5/12/2008 http://lists.oasis-open.org/archives/opencsa-liaison/200805/msg00015.htm l Approved 4. Agenda Bashing Approved 5. Issues a. Namespace for bindings and other extension points http://lists.oasis-open.org/archives/opencsa-liaison/200804/msg00010.htm l Discussion thread: http://lists.oasis-open.org/archives/opencsa-liaison/200805/msg00009.htm l Lot of discussion happened but no decisions were made. See below for the raw chat log. b. Use of Schematron in SCA Specifications http://lists.oasis-open.org/archives/opencsa-liaison/200804/msg00006.htm l Discussion on 5/5 conf-call: http://lists.oasis-open.org/archives/opencsa-liaison/200805/msg00013.htm l Did not discuss due to lack of time 6. Next Call Schedule May 29th is US holiday Jun 2nd may conflict with travel plans of members for attending SCA Assembly/Policy F2F meetings Next conf-call schedule to be discussed via email 7. AOB None ----------------------------- Raw chat log: anish: i see the two as orgthogonal anish: the two being compatibility and NS Mike Edwards: These questions alone drive me to favour as few namespaces as possible Mike Edwards: anything else rapidly becomes a nightmare Martin C: i think last week we said no more then 5 namespaces (ish) would be a nightmare Mike Edwards: > 1 is already a problem Mike Edwards: assuming they ever get used together Martin C: 1 namespace is a problem Mike Edwards: ;-) Martin C: for evolution anish: even if we move in lock step, we'll likely get > 1 namspace and will have to deal with it Martin C: agreed Mike Edwards: Just look at WSDL & BPEL - very few revisions and even they cause problems anish: sanjay, your proposal is interesting (what to do before 1.1 -- keep the same NS regardless of what changes we make, these are interim revisions) anish: i remember, during the schema 1.1 days, the W3C CR, PR and Recommendation had 3 different namespaces and it was a disaster, till all the tools moved to the Recommendation Sanjay: yes, the interim versions would just be work-in-progress in that approach Sanjay: the assumption being that we (OpenCSA MS) finish our work in reasonable timeframe Mike Edwards: Anish - it's that sort of thing that worries me a lot Mike Edwards: I really don't know who we are helping here anish: mike, right, that is why i think sanjay's proposal is promising anish: ... we say that interim versions are just that Sanjay: the early implementors can clarify their support by pointing to concrete artifacts (e.g. schema files) anish: ... no change in NS regardless of whether the changes in the spec are compatible or not anish: so, regardless of whether we have fine grained or lock-step, we have one namespace for assembly 1.1 through out it's life Mike Edwards: yes, everything before 1.1 final publication could potentially break stuff from month to month (from CD to CD) Mike Edwards: it's not nice, but trying to provide this "changing namespace" solution is a bigger nightmare anish: if we move lock-step then one NS for all specs for 1.1. If we don't move lock-step then some 3-5 namespaces for 1.1 Mike Edwards: I think that we should aim to go "final" with 1 namespace anish: so again, i think there are two separate issues: compatibility and whether we rev NS for versions before 1.1 and lock-step v. fine-grained anish: i said before that we make compatibility stmt, i think maybe we need to do this only for different final versions (1.1, 1.2, 2.0 ...) Martin C: im not sure where/why substitution groups got introduced anish: mike r, i think u are right wrt SG, but I think that is a general problem with SG and why maybe it is not a good idea to use them anish: substitution group's type is a QName not a list of QNames, so it has to be only one Sanjay: michaleR: all OpenCSA TCs use the common namespace and use fine grained namespaces post 1.1 anish: scdl has a nice ring to it Sanjay: For elements used in SCDL file, all SCA TCs use the common namespace and use fine grained namespaces post 1.1 Sanjay: s: MikeEdwards anish we're not going to have time to finish this Mike Edwards: we're going to have to table this discussion Sanjay: amendment by SimonNash: Whenever an incompatible change is made to the schema, a new rev of namespace is to be generated Sanjay: seconded by Martin Meeting adjourned due to running out of time.