W3C

Web Services Resource Access Working Group Teleconference

25 Jan 2011

Agenda

See also: IRC log

Attendees

Present
Alessio Soldano, Red Hat
Bob Freund, Hitachi, Ltd.
Doug Davis, IBM
Gilbert Pilz, Oracle Corp.
Katy Warr, IBM
Li Li, Avaya Communications
Ram Jeyaraman, Microsoft Corp.
Tom Rutt, Fujitsu, Ltd.
Yves Lafon, W3C/ERCIM
Absent
Ashok Malhotra, Oracle Corp.
Asir Vedamuthu, Microsoft Corp.
Bob Natale, MITRE Corp.
David Snelling, Fujitsu, Ltd.
Fred Maciel, Hitachi, Ltd.
Jeff Mischkinsky, Oracle Corp.
Mark Little, Red Hat
Martin Chapman, Oracle Corp.
Nathan Burkhart, Microsoft Corp.
Orit Levin, Microsoft Corp.
Paul Fremantle, WSO2
Paul Nolan, IBM
Prasad Yendluri, Software AG
Vikas Varma, Software AG
Wei Jun Kong, CA
Wu Chou, Avaya Communications
Regrets
David Snelling, Fujitsu, Ltd.
Chair
Bob Freund, Hitachi, Ltd.
Scribe
Gilbert Pilz, Oracle Corp.

Contents


<trackbot> Date: 25 January 2011

<scribe> SCRIBE: gpilz

Agenda

RESOLUTION: agenda agreed to

minutes from January 11th, 2011

RESOLUTION: minutes approved

F2F meeting preparation

Bob: like to clarify what I'm thinking about
... typically there are some issues that come up as a result of implementing, issues that come up as part of testing
... we have participants that will be at the F2F and others on the phone
... would like to group issue discussions into the morning times, testing in the afternoons

<asoldano> nice idea

Bob: is that acceptable?

(no comments on phone)

scribe: as far as timings go - is start time of 9:00 ok?

Gil: 9:00 - 5:00 is fine

Bob: group dinner on 2/15?

Gil: I can scout around
... make some recommendations

Ram: Gil, is breakfast or snacks served?

Gil: same deal as all Oracle conference center events
... they put out food for breakfast and lunch in the hall

Table of Implementations

Bob: hasn't changed since the previous call
... we've had a slew of new issues

Gil: maybe we could handle the simpler issues first?
... these are all post-CR issues?

Bob: if we resolve an issue with a substantive change - we have to go back to Last Call

Issue-11625

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11625

Doug: on DeleteMetadata - it doesn't say what to do if the target metadata doesn't exist
... would like to see it have no effect - silent ignore
... no fault etc.

Gil: could you expand?

Doug: if someone wants something deleted and it is already deleted (or just not there) - they have the intended result
... why bother them with a fault?

Gil: feeling uneasy about this - the client has no idea that what they thought was true isn't

Doug: they could always check if it mattered that much to them

Bob: we've been here before - MEX doesn't support any kind of transactionality

Katy: we've decided this before - the Delete should just fail silently
... what would you do if you got a fault anyway?

Bob: any objections to accepting as a new issue

(none)

Bob: any objections to accpeting proposal for silent fail?

Gil: where's the explicit text?

Bob: in the issue

RESOLUTION: Issue-11625 resolved with proposal in http://www.w3.org/Bugs/Public/show_bug.cgi?id=11625

Issue-11697

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11697

Doug: (explains issue)

Bob: I think there is a pre-defined SOAP fault for this

RESOLUTION: Issue-11697 accepted and left open

Issue-11698

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11698

Doug: (explains)
... implies that default for @Name is "unwrapped"

Tom: doesn't option (1) require a schema change?

Doug: yes

Tom: I like option (2) better than

Bob: any objection to accepting as a new issue?

(none)

Ram: I don't have a particular comment - I just need more time to consider this

RESOLUTION: Issue-11698 accepted and left open

Issue-11703

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11703

Doug: (describes)
... also the issue of what happens if the event source notices the bad filter later (after SubscribeResponse has been sent and processed)

Tom: what is difference between that, and your disk drive going down?
... maybe you should just kill the subscription and send a SubscriptionEnd

Doug: none, but you need to be explicit about it

Ram: I need more time on this as well

RESOLUTION: Issue-11703 accepted but left open pending further investigation by Ram

Issue-11723

Doug: (describes)

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11723

Doug: (proposes adding new fault)

RESOLUTION: Issue-11723 accepted but left open pending further investigation by Ram

Issue 11724

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11724

Doug: (describes)
... proposes rewording to make clear that an absent frag has no effect (fail silent)

Tom: when you say you don't do anything - you don't respond?

Doug: you send back a valid response message - as if the Delete had succeeded

Tom: sounds right to me

RESOLUTION: Issue 11724 accepted but left open pending further investigation by Ram

Issue 11725

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11725

Doug: (describes)
... use InvalidInput fault

Gil: (confusing rant about not requiring receiver to validate everything the consumer sends it)

(back and forth about conformance philosophy)

Tom: "MAY" be ok - or qualified MUST

Gil: "... if the service detects {foo} and chooses to fault, it MUST generate the {bar} fault"

<Tom_Rutt> but from a testing point of view, the qualified MUST is no different than the MAY, from te point of view of the client

RESOLUTION: 11725 accepted but left open

Issue-11766

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11766

Doug: reword things to make it clear that you don't have to update the resource if you don't support the schema of the thing that was passed in

RESOLUTION: Issue-11766 accepted but left open

Issue-11772

Doug: @language is required (tells you what dialect you are using)
... this is very much like a filter
... @language should be optional and default to XPath
... is this a "substantive" change?

Bob: up to the WG to decide if it is a "breaking" change

Tom: changing something from mandatory doesn't sound like a break

RESOLUTION: Issue-11772 accepted and left open

Issue-11776

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11776

Doug: (describes)

RESOLUTION: Issue-11776 accepted but left open pending more investigation by Ram

Issue-11790

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11790

Doug: (describes how we screwed up)

simplest example of this is

<mex:MetadataSection @Dialect='wsdl:definitions' ...>

<wsdl:definitions xmlns:wsdl="http://www.w3c....>

Gil: (futher explanation above)

Tom: XPath has a work-around for this, but we're not using XPath
... baffled

Katy: 'wsdl:definitions' - the "wsdl" part is shorthand for "http://www.w3...."
... the prefix doesn't matter

I like <mex:Dialect Type="{nsURI}localPart" ...

<dug> Possible serialization options, if we do want to change it:

<dug> - <mex:Dialect Type="nsURI/localPart" ...

<dug> - <mex:Dialect Type="{nsURI}localPart" ...

<dug> - <mex:Dialect TypeNS="nsURI" TypeName="localPart" ...

Tom: if we don't go from changing it from QName we'll have to use @prefixMapping from one of the WSDM-related specs

RESOLUTION: Issue-11790 accepted and left open

Issue 11849

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11849

Gil: (describes)

Proposal - replace the first sentence with the following: "This REQUIRED attribute indicates the type and version of the metadata unit contained in this MetadataSection (e.g. WSDL version 1.1)."

previous: This indicates the type and version of the metadata unit contained in this MetadataSection (e.g., WSDL version 1.1).

RESOLUTION: Issue-11849 accepted and resolved with proposed text from: http://www.w3.org/Bugs/Public/show_bug.cgi?id=11849

Issue-11850

http://www.w3.org/Bugs/Public/show_bug.cgi?id=11850

Gil: (describes)

RESOLUTION: Issue-11850 accpeted and left open pending further investigation by Ram

<dug> http://www.w3.org/Bugs/Public/show_bug.cgi?id=11865

Issue 11865

Katy: I hadn't thought of this as a problem on the Get
... 2 issues - (a) Get and (b) Put/Delete
... only solution I see is Replace or Delete all the documents identified by the Dialect/ID tuple
... Get is another matter

Doug: solution in my mind is what Katy was saying
... Get and Put contain "complete" set of all metadata matching the Dialect/ID tuple
... interesting twists around @Content
... if I ask for everything that can be referenced via EPRs - clearly I might not get the "complete set"
... we need something that says "if you have multiple sections for the same Dialect/ID/Content triplet, they are all separate parts of 'the same thing' and there are no duplicate sections"

<Tom_Rutt> I am confused, if there are several schema file with the same namespace, and that namespace is used as @Identifier, would every root element in that "combinded schema" across the namespace be allowed to be send in the return? what if there is a single global element defintions which is a multiple container?

Doug: currently there is nothing in the spec that prevents you from returning a Metadata document with duplicate MetadataSections that contain the exact same XML

Tom: I'm confused - when you are talking about multiple schemas with the same targetnamespace

RESOLUTION: issue 11850 accepted and left open pending some reasonable proposal

<dug> http://www.w3.org/Bugs/Public/show_bug.cgi?id=11867

Issue-11867

Doug: (describes - editorial)

RESOLUTION: Issue-11867 accepted and resolved with proposal in http://www.w3.org/Bugs/Public/show_bug.cgi?id=11867

CR progression

Bob: text is available for review
... on home page
... need to work on indicating which features are at risk

Li> new CR version - is it stable enough to implement?

Bob: only thing that may change is "status of this document" section

Doug: if these are the docs we are going to be testing against - we need to know what the namespaces will be so we can use those

Bob: depends on the timing of when we roll out the specs
... you can use the current namespace for the Feb testing

Doug: I'll update the scenario doc to indicate this

Test Scenarios

http://www.w3.org/2002/ws/ra/edcopies/scenario.html

Bob: I've noticed a few changes going by
... how are we doing for Feb?

Ram: I think these are fairly stable and implementable

Doug: this doc needs to go through more changes - more variations
... mainline scenarios are there - but we need to add variations
... missing critical bits of info
... Eventing and Enum talk about testing *End
... but how will this be triggered?
... need to specify in the doc

Ram: as long as the core scenarios are intact we can do the fine tuning later

Bob: we're going to test, grind through, find issues (spec, scenarios, test infrastructure)
... will need to take another whack at testing

Ram: can we call "freeze" on what we have in the scenario today?

Bob: plan to test what we have now

Doug: I'm OK with the idea that we don't have major changes
... not comfortable with the idea that the scenario won't change

Bob: just want to set expectations about what people should bring to the table in Feb
... as firm as a marshmallow - but formed
... need to give devs something to shoot at

Doug: people shouldn't be surprised if additional scenarios/cases are added

Ram: I'm ok with that - as long as we have a stable version to shoot for

post CR

Bob: we need to declare a date that defines how long the CR will last (minimum review period)
... can't go to PR before that date

Tom: if we have some of the schema changes it looks like we might have
... can we progress to PR, even if we change the namespace?

Bob: if we make a "breaking change" (lots of changes are not breaking)

Tom: we can have a new namespace for the PR schema from the CR schema

Bob: you don't necessarily have to change the schema

All: (reviews proposed directions to issues and concludes that we will probably have to change the namespace)

Bob: we may need a 2nd interop
... April 29th end of review period

<dug> 60 days for me

<dug> so march

<Bob> march 31

RESOLUTION: end of CR period March 31st, 2011

<Bob> rrsagent generate minutes

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2011-02-05 23:14:34 $