WS Description WG telcon
15 Jun 2006

See also: IRC log


Allen Brookes, Rogue Wave Software
Roberto Chinnici, Sun Microsystems
Glen Daniels, Sonic Software
Paul Downey, British Telecommunications
Tom Jordahl, Macromedia
Jacek Kopecky, DERI Innsbruck at the Leopold-Franzens-Universitšt Innsbruck, Austria
Amelia Lewis, TIBCO
Philippe Le Hegaret, W3C
Jonathan Marsh, Co-chair/Microsoft
Jean-Jacques Moreau, Canon
Gilbert Pilz, BEA Systems
Tony Rogers, Co-chair/Computer Associates
Arthur Ryman, IBM
Charlton Baretto, Adobe Systems
Youenn Fablet, Canon
Vivek Pandey, Sun Microsystems
TonyR, Jacek


Implementer's call

<TonyR> http://lists.w3.org/Archives/Public/www-ws-desc/2006Jun/att-0033/20060608-ws-desc-minutes.html

<TonyR> SCRIBE: TonyR

Action Items

Chairs to invite XFire: done - Jonathan invited them

Jonathan to investigate SOAP 1.2 + .Net: done

Arthur to add wrappers to the schema: done

Arthur to put parent, feature, properties to the base schema: done

Arthur: we should add an {extensions} property to the component model to list the extensions supported by the processor, because we need this to clarify why two component models might differ given the same input documents

Jonathan to publicise interop: done

Arthur takes responsibility for the testcase for unknown extension with wsdl:required false

Arthur takes responsibility for the testcase for unknown extension with wsdl:required true


Arthur: Axis has reported that they have complete integration with Woden
... has written an visa support letter for an attendee from India
... need to build a validation report (used for comparing with the test metadata, for comparing the assertion)

<scribe> ACTION: Jonathan to build a new XSLT to construct the validation reports [recorded in http://www.w3.org/2006/06/15-ws-desc-minutes.html#action01]

<lmandel> +1.416.95.aaaa = lmandel

<JacekK> scribe: Jacek

<JacekK> scribenick: JacekK

action item review

3.  Review of Action items [.1]. 

DONE      2006-05-11: [interop] Chairs to invite XFire to join this 
                      call per Tom Jordahl's suggestion 
RETIRED   2006-05-11: [interop] Jonathan to look into how to create 
                      SOAP 1.2 services with .NET. 
DONE      2006-06-01: [interop] Arthur to add wrappers to the schema.
DONE      2006-06-01: [interop] Arthur to put parent, features,
                      properties in the base schema.
?         2006-06-01: [interop] Jonathan to add sorting of soap modules,
                      http/soap headers.
?         2006-06-01: [interop] John to file issue whether {rpc
                      signature} should be OPTIONAL (4.1.1).
DONE      2006-06-01: [interop] Jonathan to publicize interop event.
?         2006-06-08: [interop] Arthur to create a testcase for 
                      an unknown extension wsdl:required=true.
?         2006-06-08: [interop] Arthur to create a testcase for 
                      an unknown extension wsdl:required=false.
?         2006-06-08: [interop] Arthur to write test cases for messages 
                      and message exchanges for a simple WSDL (eg: an 
                      echo web service).

?         2005-07-21: Pauld to write a proposal for a working group 
                      report for requirements for schema evolution 
                      following closure of LC124 
?         2006-03-30: Marsh to make XSLT improvements for RDF
?         2006-04-20: Glen to flesh out a model for runtime test
?         2006-06-08: Charlton to review the text appropriate to CR44 
                      to ensure that it is adequately explanatory.
[.3] DONE 2006-06-08: Paul to address the issues in CR45 regarding 
                      schema without namespace.
?         2006-06-08: Gil to write a response to the raiser of CR47.
DONE      2006-06-08: JacekK to respond to Jonathan re: cf050.
DONE      2006-06-08: Jacek to write a clarification to Arthur's 
                      proposal, explaining that BindingFault is correct,
                      not BindingFaultReference.

Current Editorial Action Items
 - none -

Note: Editorial AIs associated with LC issues recorded at [.2]. 

[.1] http://www.w3.org/2002/ws/desc/#actions
[.2] http://www.w3.org/2002/ws/desc/5/cr-issues/actions_owner.html
[.3] http://lists.w3.org/Archives/Public/www-ws-desc/2006Jun/0025.html

Glen: (re his action) I'll make a serious attempt to flush that out by the end of the week

Jonathan: the Toronto meeting is coming up soon


next meeting: June 29

<Arthur> coming

will be chaired by Tony

Jonathan: should we have telcon in the interop event week?

Arthur: 42
... the call could be useful to update the WG on progress

Jonathan: so we'll only dropping 06/22 call
... wrt the interop event, we seem to be pretty much set, Glen please register

Glen: I'll only come if I get time before to work on the impl

Jonathan: interop TF meeting talked about a group of issues that need to be resolved
... it's CR050 - an optional extension has required property with default value
... question about when it's in the CM

<Arthur> http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/test-suite/interchange/wsdlcm.xsd?rev=1.8&content-type=text/xml

Arthur: we need to be explicit about what extensions a component model claims to know and support
... we can't know what extensions are in effect by looking at a WSDL document

<Arthur> MUST occur if the rpc extension is

<Arthur> supported AND the {style} property of

<Arthur> the Interface Operation component

<Arthur> contains

<Arthur> http://www.w3.org/2006/01/wsdl/style/rpc.

Arthur: (goes through particular occurrence constraints of extension properties)

Glen: so the property {style} would exist, and then the particular extension properties

Jonathan: so the rpc signature property doesn't have to appear when an operation doesn't follow the rpc style

Arthur: I'm proposing we add {extensions} property to Description component
... presence of an extension property then depends on the extension being present in the {extensions} property, and possibly on other conditions as well
... 4.1: the RPC style is selected by including the URI in the style property

Jonathan: but does that imply that the signature property has to be there?

Arthur: that's the clarification:
... for the RPC extension, the co-occurrence constraint is the existence of the RPC style on that operation
... in other words, the property is required, but we're clarifying when it is required
... we may require that rpc:signature attribute would be required

Glen: signature can be implied from the schema

Arthur: in that case, we are defining a default value

Jonathan: so the {rpc signature} prop could be optional?

<Arthur> http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/wsdl20/wsdl20-adjuncts.html?content-type=text/html;%20charset=utf-8#RPCStyle

Glen: we should have an example there
... but we have all examples in the primer

Roberto: the {rpc signature} should be optional

Arthur: we can say here that it's optional, there will be more required extension properties

TonyR: not completely happy about making this optional, the information should be required

Glen: you can derive the information from the schema; we don't have rules for doing that

Jonathan: we may not need such rules, the user is free to interpret the operation as they wish

Glen: the rules could be useful

Jonathan: but the client doesn't have to use the RPC signature

Arthur: if the service provider wants to suggest a signature, they will provide it

Jonathan: it sounds like a different issue, maybe Glen should raise it
... trying to do minimal changes to the spec to make it correct and consistent

Glen: I prefer required property with algorithm for defaulting

Jonathan: there seem to be a few people for optional, a few for required

Arthur: we should request a concrete proposal for the defaulting algorithm in a timely basis

Glen: let's make it optional now, maybe a concrete algorithm can be proposed later

Jonathan: I'll add this issue (and send the email), propose that it become optional, others may counterpropose defaulting rules

<Jonathan> ACTION: File this issue. [recorded in http://www.w3.org/2006/06/15-ws-desc-minutes.html#action02]

<pauld> wonders what the algorithm would be - do people care if one tool generates "bar echo(bar) " as opposed to "void echo(ref bar)"

Jonathan: so if it's optional, it's only contributed when the extension appears
... any other preconditions?
... if I have rpc:signature but the style doesn't contain rpc style?

Arthur: that'd be an error

JacekK: if somebody adds new style that's RPC-like, they couldn't reuse our attribute

Jonathan: their spec could add that reuse (re-enable it)
... I'll add the options to the issue

<Arthur> MUST occur if the wsdlx extension is

<Arthur> supported.

Arthur: then there is the safety property

Roberto: the extension is wsdlx:safe, not wsdlx

Arthur: I propose that extensions are identified with namespaces

<Arthur> CR 050 http://lists.w3.org/Archives/Public/www-ws-desc/2006Jun/0034.html

Jonathan: so you're proposing the {extensions} property for Description component
... why does it need to be in the component model?

Arthur: we're trying to solve question about validity of an instance of component model
... we need to know what extensions (by default optional) are in effect

Roberto: but validity can be function of two arguments - component model + set of extensions

Arthur: I don't see adding {extensions} to the CM as artificial
... you also embed DTDs, namespaces

Roberto: can we look at extension model of XSLT?
... extensions not necessarily tied to namespaces

JacekK: maybe let's just add it to the interchange format, not to the component model

GlenD: this matters when reading the syntax and building component model; and when checking validity of component model
... we're talking about the latter here

Jonathan: yes, we don't have extension requiredness in the CM either

Arthur: re: extensions and namespaces
... in W3C things are named with IRIs, extension spec should have an IRI
... it's natural that the URI would be used as the namespace
... but the URI need not be the namespace

Jonathan: the component model is a "platonic ideal", adding this info violates that
... any given implementation can't know *all* the extensions, it will have a limited view
... if I know extension foo and you don't and it's not used, we may have trouble in comparing CMs

JacekK: the issue came up for testing, we can only say for our CR testing what extensions should be supported for it

Arthur: WSDL2 is extensible, it will have extensions, we also need to test the general mechanism

<Zakim> JacekK, you wanted to talk about role of interchange format

JacekK: the component model interchange format is only for the CR testing

Arthur: there were ambiguities discovered

JacekK: and we resolve the ambiguities and tell the buggy implementations that they are buggy

Jonathan: can we resolve that by adding the {extensions} property?
... and extensions authors should be clear about when their properties appear

Arthur: my proposal makes it very explicit when properties appear

Jonathan: it doesn't seem that adding the property actually solves our problem

JacekK: we don't have the URIs currently

Arthur: we do have 4 extensions and 4 uris, it would work

GlenD: we have 2 separate issues here: I think it's useful for extensions to have URIs; but I don't think it's necessary to have the list of support extensions in the component model

Jonathan: we can add a general prose in the core, but I'm unhappy about the property

GlenD: if I know 200 extensions, do I mention all of them?

Arthur: yes

GlenD: can we make this more efficient, only mention those used?

Jonathan: but a WSDL file doesn't have to use an extension for it to show itself in the component model in one parser

GlenD: lemme repeat: if my processor understands addressing, when it reads any WSDL, it will decorate operations with actions
... seems odd to me

Jonathan: in CR22 we're trying to address an alternative - re-parsing all WSDL files on import

GlenD: is this a processor-specific optimization?

Arthur: when do you add safety property?


Arthur: I consider it a bad form to have optional extensions have required properties or even decorate components that don't get extension markup, but we have examples of that out there
... in practice, every extension will probably have markup, so they will have namespace, we can use it

JacekK: maybe let's rename the URI .../wsdl-extensions to ../wsdl-safety, say all extension have to have URIs, then we'll have them
... namespaces themselves are not quite good enough

Jonathan: is this only for component model comparison?

Arthur: also validation

Roberto: arthur says "given a component model, is it valid?", I'd say "given a component model and a set of extensions, is it valid?"
... only when you bring the info into the component model, you need actual URIs

Arthur: there are two designs, yes: component model and a bunch of specs; or you put the bunch of specs in the component model
... the component interchange format would benefit

<Zakim> JacekK, you wanted to say component interchange format doesn't need it

JacekK: can we just extend the component model interchange format for our purpose here?

Jonathan: we can even only say "let all our extensions be supported"
... back to CR22
... proposal: a WSDL file should be parsed independent on where it's imported

GlenD: I'm curious about the actual spec wording of this
... this might touch interestingly on how the processors work with WSDL files

Arthur: you need some considerations in the language design against inordinate computing burdens
... none of the extensions we have so far would be affected

Jonathan: WS-A would also work for which we are fairly happy

<Jonathan> ACTION: Arthur to update CR022 proposal. [recorded in http://www.w3.org/2006/06/15-ws-desc-minutes.html#action03]

Arthur: every required property in extensions needs to say when it's required

Jonathan: I feel we could add some text to part 1 on that

Arthur: I can try to do that

<Jonathan> ACTION: Arthur to propose part 1 test about REQUIRED extensions. [recorded in http://www.w3.org/2006/06/15-ws-desc-minutes.html#action04]

Arthur: another specific problem - when are HTTP and SOAP bindings (extensions) present?

Jonathan: yes, that's CR55
... I'll try to consolidate the issues for next time

Meeting adjourned

implementors' call will still happen next week

Summary of Action Items

[NEW] ACTION: Arthur to propose part 1 test about REQUIRED extensions. [recorded in http://www.w3.org/2006/06/15-ws-desc-minutes.html#action04]
[NEW] ACTION: Arthur to update CR022 proposal. [recorded in http://www.w3.org/2006/06/15-ws-desc-minutes.html#action03]
[NEW] ACTION: File this issue. [recorded in http://www.w3.org/2006/06/15-ws-desc-minutes.html#action02]
[NEW] ACTION: Jonathan to build a new XSLT to construct the validation reports [recorded in http://www.w3.org/2006/06/15-ws-desc-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2006/06/15 16:35:29 $