Web Services Resource Access Working Group Teleconference

05 May 2009


See also: IRC log


Ashok Malhotra, Oracle Corp.
Asir Vedamuthu, Microsoft Corp.
Bob Freund, Hitachi, Ltd.
David Snelling, Fujitsu, Ltd.
Doug Davis, IBM
Fred Maciel, Hitachi, Ltd.
Geoff Bullen, Microsoft Corp.
Gilbert Pilz, Oracle Corp.
Katy Warr, IBM
Li Li, Avaya Communications
Jeff Mischkinsky, Oracle Corp.
Tom Rutt, Fujitsu, Ltd.
Vikas Varma, Software AG
Wu Chou, Avaya Communications
Yves Lafon, W3C/ERCIM
Bob Natale, MITRE Corp.
Mark Little, Red Hat
Paul Fremantle, WSO2
Prasad Yendluri, Software AG
Ram Jeyaraman, Microsoft Corp.
Ranga Reddy Makireddy, CA
Sreedhara Narayanaswamy, CA
Sumeet Vij, Software AG
Bob Freund, Hitachi, Ltd.
Gilbert Pilz


<trackbot> Date: 05 May 2009

<Bob> Agenda: http://lists.w3.org/Archives/Public/public-ws-resource-access/2009May/0006.html

<Bob> scribe: Gilbert Pilz

agenda bashing

RESOLUTION: agenda approved by UC

approval of minutes from 04/28/2009

RESOLUTION: minutes approved by UC

F2F scheduling

Bob: June 9-11 at Oracle HQ in Redwood Shores, CA
...: get your RSPs 2 weeks prior or you won't be fed
... I will set up a ballot that will close on 5/26/2009
... we are planning on all three days

new snapshots

Bob: we've close a few issues so we should take a new snapshot
...: Yves, can you take care of that?

Yves: yes

<Yves> ACTION: Yves to generate may 1st snapshots [recorded in http://www.w3.org/2009/05/05-ws-ra-minutes.html#action01]

<trackbot> Created ACTION-64 - Generate may 1st snapshots [on Yves Lafon - due 2009-05-12].

Bob: just review the things that have closed since the last snapshot
...: the W3C process specifies a 3 month heartbeat
... on that basis, we should have a new WGD out in June
... Do we have enough time to put togther such a thing, and approve it at the F2F?

Wu: Will this WGD be the final one?

Bob: There is a WGD that occurs every 3 months. There is no progressive status. It is not last call or anything like that.

Wu: OK

Gil: Want me to create an AI?

Bob: Too early. Wait until the end of May snapshot.

AI review

Bob: Would every one who has an AI please review them and either (a) adjust the date or (b) offer it up for someone else to do.

New Issues

Bob: One new issue, Gil's, 6860
...: any objection to accepting this and assigning to Gil?

RESOLUTION: 6860 accepted and assigned to Gil

Task Team Progress

Bob: Who will speak for team 6413?

Geoff: Not really clear what the scope of deliverables is.
...: Microsoft and IBM are working on a list of questions to be answered, but there are two main points of contention.
... We can't make progress until we resolve these.
... Should the fragment specifiers be in the same namespace as WS-T?
... Should the fragment dialects, operations, etc. be in the same specification as WS-T?
... It's time to return to the WG on how to decide these issues.

Bob: Do other member of the Task Team concur?

Katy: Yes.

Bob: It seems that these issues are described. Are they described enough to deal with them on this call?
... Does everyone have their arguments in place?

Katy: Yes.

Bob: Should we deal with them in that order, or the opposite order?
... They are kind of connected, right?

Geoff: Yes.

Bob: Sould the fragment specification, whever it is contained, be in the WS-T namespace or not?

Katy: We discussed this with our dev teams.
... Our products teams would not implement WS-T without at least basic fragment support.
... The reason that WS-T and WS-RT were split is primarily historical.
... The key reason why it is bad to split these into two namespaces is that it allows a loophole of different fragment mechanisms within WS-T.
... Putting them in the same namespace/spec removes any ambiguity about how fragments should be supported in WS-T.

<DaveS> I strongly support Katy's notion that one namespace supports better chances of interoperability.

Bob: (missed question)

<dug> Bob: can frag support be used outside of T?

<Ashok> Yes, I also support Katy's notion that one namespace supports better chances of interoperability.

Katy: We need to look at the 99% use case. We've looked at splitting it out into an appendix.

<dug> Katy: maybe in theory but we don't see it in practice

Bob: Any further arguments in favor?
... Any against?

Geoff: Not sure where to start here . . .
... One of the primary reasons for putting it in the WS-T namespace seems to be to force people to adopt something.
... This seems very strange. You're not going to be able to do this.
... We don't have the right to do this.
... We can offer up specs and, if we do things right, people will use them.
... But it's not right to force people to do things they don't want to do.
... Other issues: no real reason to make this change.
... No evidence that I have seen that this is other than political
... These have been separate in the past, they work separately, there's no value to combining them.
... WS-T has been widely adopted, widely referenced, and widely implmented.
... There are reasons for combining specs, we should stick to those.

Bob: Will allow Katy to rebut.

Katy: This isn't forcing anyone to do anything. It simply specifies the recommended way of supporting fragment access.
... This has nothing to do with politicis. It is simply about support fragment access.
... It also provides a pattern for fragment extensibility that other specifications can use.

Geoff: We disagree.
... I don't believe that there are any serious technical reasons for doing this.

<dug> constrained devices (which are impacted by additional namespace) would be impacted by having to support 2 specs instead of 1- and they would use frag support exactly because they're constrained.

Geoff: This is just joining things together "for advantage".

Wu: I'm kind of confused and concerned.
... The group has already reached consensus to split the two (WS-T and frag)
... This has already been decided.

<asir> Yes the WG reached consensus prior to forming 6413 taskforce

Wu: This appears to be a step back from the previous consensus.

Yves: I want to restate what I said at a previous concall.
...: For me, fragments are really part of the EPR
... The definition of fragments is in the ??? spec
... The meaning of the fragment is usually related to the MIME type
... In this case things are different because we constrained to XML?

<Geoff> +1 to Yves
...: Don't see a reason to combine them.

DaveS: Primary reason to put fragments in WS-T is to promote interoperability.

<Geoff> interop can be achieved without joinign them together

DaveS: I agree with Geoff that the only way this will work is if we produce a high-quality spec that is easy to use.

<Yves> as long as a namespace is chosen for fragment, it will provide interop, there is no need to group specs on namespaces (apart to reduce the number of ns URIs in implementations runtime)

DaveS: We need to produce one spec to pull the community together.

<dug> Gil: to Wu (re: this was decided) - not my understanding

<dug> ... the task force was to look at what things would look like _if_ there was a frag spec

<dug> ... it was not cast in stone - examine a possibility

Wu: disagrees
... I remember it being a consensus

Bob: My recollection was that the idea that it could be a separate specification illicted less negative reactions than the inclusion idea.

<asir> Sounds like we should think about documenting the starting point and the goals and questions for any future taskforces

<Geoff> thankyou Bob

Doug: I agree with Gil's interpretation
... IBM think a separate frag spec will not work because a reference from WS-T to another spec isn't a strong enough link to assure interop.
... One primary example is Member Submission specs.
... They all say "THis doc is just for evaluation" "This may change"
... People ignored that.
... People implemented these specs and now scream loudly when you propose changes.
... So its unlikely that a simple reference and admonishment will produce the desired result (interop)

<Geoff> W3C has plenty of precedences for establishing such links - SOAP, WS-A and WSDL

Doug: To Yves, I'd like to see a concrete proposal around your ideas.
... I can't see how that would work.

<asir> W3C has plenty of precedences for establishing such links - SOAP, WS-A and WSDL

Doug: Fragment support is driven by the client and service-minted EPRs should not be changed by the client.

Yves: The thing is . . .
... In a way you are using the EPR if you use the specified policy to talk to the service.
... You should be able to say that you only want "one part" of "this" (the thing referenced by the EPR)
... This should be part of WS-Addr so you don't need to tie it to WS-T
... It should support more that just WS-T
... In a way you are right about not changing EPRs . . .

Geoff: There is plenty of precendent in the W3C for having links between specifications (SOAP, WS-Addr)
... This is a commonplace way of doing things.
... Yves comment is interesting and valid.
... WS-Man was thinking that way
... They put the definition of the resource in the header, not the body of the message.

<jeffm> note: SHOULD is a meaningless way of saying "please, but if you don't, oh well"

Geoff: They were working towards this idea of the concept of a frag being an extension of the concept of an address.

Bob: Do we have a consensus that frags do need to part of our work?

Doug: It is part of our charter.

Bob: Nobody disputes we need frags as part of our output.
... The question is, where do we put it?
... In the input specs, it is part of WS-RT.
... We could just mumble along and have it part of a WS-RT that we produce.
... Thinking about our alternatives, the namespace issue connects frags to WS-T.
... Even if frags were in WS-RT, a common namespace would connect them.
... Conceptually writing it as an EPR extension is another approach.
... catalog approaches
... (a) Include (in WS-T with same namespace, (b) Separate (with WS-T namespace), (c) Separate (as an EPR)?
... (c) do nothing

<Yves> isn't "do nothing" a (d) ?

<asir> what is do nothing

Gil: It's a 3 x 2; in WS-T, in separate spec, as an EPR extension then in/out of the WS-T namespace - then "do nothing"

<dug> a) in T, b) in separate RT spec, c) EPR extension ?

Gil: (a) in WS-T, same namespace as WS-T
... (b) in a separate spec, same namespace as WS-T

<dug> asir: that's the msft motto

Gil: (c) as an EPR extension, same namespace as WS-T

<dug> :-)

Gil: (d) in WS-T, different namespace than WS-T
... (e) as a separate spec, different namespace than WS-T
... (f) as an EPR extension, different namespace than WS-T
... (g) do nothing (WS-RT as today)

<asir> Thanks for clarifying 'do nothing'

(all) Ridicule Gil for his noisy keyboard

Bob: We don't have Chad, but we do have Open STV

Jeff: Why don't we have Chad?

Bob: Yves?

Yves: We'd have to get it from Paul Downey

Jeff: Open STV works just fine

Bob: Think about your choices, type the letters 'a' through 'g' in IRC scrambled to your taste
... First one most favorite, last one least.

Jeff: As many as you want

<Geoff> e, g, f,

<dug> a,d,b,g

<Yves> f e g d

<li> f,e,g

<DaveS> Fujitsu: a, b, d, e, g

<Katy> a d b g

<Wu> egf

<Ashok> a,b

<Vikas> a,g

<asir> egf

<TomRutt> a, b, d, e, g

b, a, d, e

<jeffm> a,b,c

<asir> what about Bob?

<Geoff> I assume this is a straw poll, right?

(all): Attempt to work Open STV

Geoff: Just to be clear, the purpose of this poll is to winnow things down?

Bob: Yes.
... We should have done a dry-run of this before . . .
... results
... a - 7
... e - 3
... f - 2
... b - 1
... (a) wins on the count of first choices
... (a) was in same spec as WS-T with the same namespace as WS-T

<Bob> Election title: ballot 1

<Bob> Method: Scottish STV

<Bob> Number of total ballots: 13

<Bob> Number of invalid or empty ballots: 0

<Bob> Number of ballots used in the count: 13

<Bob> 7 candidate running for 1 seats.

<Bob> R|a |b |c |d |e |f |g |Exhaust|Surplus|Thresho

<Bob> | | | | | | | |ed | |ld

<Bob> ==================================================================================================================

<Bob> 1|7.00000|1.00000|0.00000|0.00000|3.00000|2.00000|0.00000|0.00000|0.00000|7.00000

<Bob> |---------------------------------------------------------------------------------------------------------------

<Bob> | Count of first choices. Candidate a has reached the threshold and is elected. The election is over since all

<Bob> | seats are filled.

<Bob> Winner is a.

Bob: To take it from here, things other than 'a' or 'e' seem to have little support.
... Hitachi doesn't have that strong an opinion on this
... 'a' seems to have the advantage by more than a factor of 2
... For those people that support things other than 'a', is there a way 'a' can be made palatable.

Asir: What was the question?
... Alternatives other than 'a' don't have enough support that, even if they coalesced, couldn't win out over 'a'.
... This is just a strawpoll, right?

Jeff: This clearly shows that there is a group that supports 'a' and a group that supports 'e'. Other than that, there isn't much support for the others.

<asir> fruitful options are a and e

Bob: The only two that are fruitfull are 'a' and 'e'.
... We can eliminate the others

DaveS: If we do 'e', it is a separate frag spec, not the full WS-RT

Bob: Agrees.
... But its not clear at that point what would remain WS-RT

Asir: We have an issue on this.

Bob: For those people that want to move frag support, what would be left in WS-RT.

Doug: Mulitple frags, metadata on ???, resource lifecycle stuff

DaveS: multiple dialects

Asir: these need to be documented

Doug: They are in the WS-RT spec.

Bob: Trying to find another alternative . . . we've come down to a pretty black and white situation.

???: looks like it Bob.

Bob: Anyone object to taking this to a decision?

Geoff: What do you mean "a decision"?

Bob: Another poll (binding?)

Asir: I think we can explore other alternatives.

Bob: Do you have one?

Asir: Not off the top of my head.

<DaveS> I'd like to go for a decision.

Jeff: How many months would you like to come up with another alternative?

(Asir & Jeff): {discuss the amount of time spent on this issue}

(Asir & Bob): {discuss the amount of time spent on this issue}

Bob: How much more time are we going to spend looking for an alternative? There don't seem to be any.

Asir: Not right now.

DaveS: There have been a lot of ideas and proposals.

<Yves> I wonder if the TAG would have an answer for this one

DaveS: Nothing seems to be gaining traction.

Bob: Yves, are you asking for a TAG ruling on this one?

Yves: We can always ask.

<asir> We could ask the TAG a specific question ..

Ashok: Is this a technical question or a procedural question?

<asir> re teh topic

Ashok: The TAG only does technical questions.
... What are you going to ask them?

<Yves> the question would be "where does a fragment definition belong, addressing or actions"

Bob: What I'd like to do at this point is to push this off until the next call.

<asir> Yves: that is a good question to ask

Bob: The problem with the polarization is that the fundamental point revolves around the namespace.
... There doesn't seem to be any middle ground.
... We've reduced the issue to a very binary, polarizing choice.
... I'm going to call around to see if there are any reasonable alternatives.
... If we don't have what looks like the start of a reasonable alternative, we're just going to have to decide.

RESOLUTION: issue defered until next week

team 6401

Wu: we have a concrete detailed proposal based on Gil's approach
...: We sent this to Gil and Geoff
... We are awaiting feedback from Gil

Bob: When did you send this out?

Wu: Last Monday

<DaveS> +1 to Ashok.

Asir: Should we use the public mailing list?

Ashok: Yes

Gil: I got Wu's proposal this Monday (5/5/2009)

Wu: Would you like me to send this to the group list?

Bob: Yes

Wu: We'd like to work this out before brining it to the group, but OK.

<dug> LOL another easy one


Gil: there are 4 proposals on the Wiki
... they look like valid candidates

Geoff: I wasn't aware that we were using the Wiki

Bob & Gil: {comment on desireable qualities of the Wiki}

Geoff: but some of the pros and cons seem pretty subjective
... what do you do if you don't agree?

Bob: you can edit them
... The whole idea was simple to get the list of proposals

Geoff: that's what I thought
... but they have "pros" and "cons" associated with them

Bob: You can delete them

Doug: or add to them

(all): {discussion about process of using Wiki}

Bob: I do want to get to a list that we agree is a minimized list of options.

<asir> Asir said, re 6413 strawpoll ... if we purge duplicate reps in the poll then the result is (a) is 4 and (e) is 3

<Bob> below was inserted in the minutes to reference the actual STV results after the meeting from the data collected during the meeting

<Bob> I posted two election results of STV (by person) at http://lists.w3.org/Archives/Public/public-ws-resource-access/2009May/0014.html result was a & e after 6 rounds

<Bob> and two election results (By company) at http://lists.w3.org/Archives/Public/public-ws-resource-access/2009May/0016.html result was a & e after 5 rounds)

<Bob> and possibly what would happen if one more company, say Hitachi, voted at http://lists.w3.org/Archives/Public/public-ws-resource-access/2009May/0015.html result was a & e after 6 rounds

<Bob> The only thing I think you can draw from this is that a & e are the primary alternatives

<Yves> trackbot, end teleconference

Summary of Action Items

[NEW] ACTION: Yves to generate may 1st snapshots [recorded in http://www.w3.org/2009/05/05-ws-ra-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2009/05/13 07:47:33 $