See also: IRC log
Present:
Erik Ackerman Lexmark Allen Brookes Rogue Wave Software Roberto Chinnici Sun Microsystems Glen Daniels Macromedia Youenn Fablet Canon Dietmar Gaertner Software AG Martin Gudgin Microsoft Jacek Kopecky Systinet Sandeep Kumar Cisco Systems Philippe Le Hégaret W3C Amelia Lewis TIBCO Kevin Canyang Liu SAP Lily Liu webMethods Jonathan Marsh Chair (Microsoft) Dale Moberg Cyclone Commerce Jean-Jacques Moreau Canon Don Mullen Tibco Arthur Ryman IBM Adi Sakala IONA Technologies Jeffrey Schlimmer Microsoft Jerry Thrasher Lexmark Steve Tuecke Global Grid Forum William Vambenepe Hewlett-Packard Sanjiva Weerawarana IBM Steven White SeeBeyond Umit Yalcinalp Oracle
Regrets:
David Booth W3C Steve Graham Global Grid Forum Tom Jordahl Macromedia Steve Lind AT&T Barbara Zengler DaimlerChrysler Research and Technology
Chair: Jonathan Marsh
Scribe: sanjiva
DONE [.2] 2003-01-16: Sanjiva to explain why naming faults is
unnecessary
? 2003-01-21: Roberto and gudge to create a branch and work up
a binding proposal based on referencing type
systems directly from operation components.
(Umit's example, Sanjiva's example, WSDL 1.1
example, and others.)
? 2003-01-30: Jacek to write up text on SOAP response MEP after
Gudge and Jeffrey send their proposal for
request/response MEP.
DONE {.4] 2003-01-30: Gudge to check that we have an issue regarding
being able to specify the verb on a per
operation basis.
DONE [.5] 2003-02-06: Jeffrey to synchronize the definitions by
referencing the terms in the glossary.
DONE [.3] 2003-02-06: Jonathan to send a summary about HTTP binding
issues.
<Philippe> Wednesday session: http://lists.w3.org/Archives/Member/w3c-ws-desc/2003Jan/0003.html
Jonathan: remember to register for the F2F
JM: Overall goal for F2F is to close remaining outstanding stuff for Part I
<jeffsch> http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/issues/wsd-issues.xml#x53
JM: Gudge suggests we go to last call for part I if we get all the topics covered
... properties & features TF is meeting on Tuesdays at this time.
... We should consider dropping the use case / scenarios doc if we are not willing to put the necessary time and effort on it. WS-Arch doesn't appear to be contributing as we expected.
SK: Explains what DaveO did to get some use-cases for WS-Arch.
... Not many comments from either group on the doc - so DaveO also put it on the back burner. Apparently DaveO is willing to help if people are interested in doing the use-cases doc.
JM: Is concerned that getting a doc thru to finality requires a lot of grunt work that the editorial team does not appear to be willing and able to do at this time.
... The May F2F will require remote participants to be on IRC in order to be on the phone; that makes it much easier for the host to do the calling and to know when to call etc..
... New issue: WSDl 1.1 bugs; not our problem but we should make sure we fix in 1.2.
... New issue: Mark Baker's note about app protocols .. not sure what to do about that yet.
... Topic: Operation Name Uniqueness
<JacekK> Sanjiva: I've proposed that we might bring back overloading
... Sanjiva: nobody has demonstrated why we shouldn't do this
... JM: we've removed overloading a long time ago, we might not want to reopen it
... SW: it was me who wanted to remove overloading in the first place because noone was using it, this has apparently changed
JM: Would like to take a straw-poll
Roberto: Agrees that proposal is not that elegant, but it gets the job done. Thinks that in the grid scenario users aren't overloading intentionally but rather accidental overloading.
Scribe: Steve Tuecke (ST): GGF position is that status quo prior to F2F is totally unacceptable. The F2F propsoal is minimally acceptable because it solves the accidental collision problem.
ST: Roberto commented that purposeful overloading is not common. Disagrees - GGF would like purposeful overloading.
... GGF is moving towards document centric interfaces. Common pattern is to create a base interface that defines an operation with a particular parameter with a particular purpose. Would like to refine in a sub-interface and further type that parameter.
... E.g., go from a compound any to a fully typed thing.
... Would prefer real overloading, but will have to live without if necessary.
Scribe: Amy Lewis (AL): Thinks overloading is an enormous topic and is concerned WSDL can take it on without defining exactly what is meant by overloading.
AL: Language bindings can be complicated: e.g., automatic promotion in certain languages.
<JacekK> Sanjiva: WSDL is not a programming-language-level thing so while I agree we'd have to make some compromises on overloading it wouldn't be so much problem
Gudge: At a portType level WSDL is about messages that go in and out.
... thinks that using programming language usage is not what WSDL is about.
JM: Wants to take a straw poll to see what to do (doesn't hear anything new).
Scribe: OOops.. thanks gudge ;-)
<Gudge> no worries
<Marsh> Straw poll: 1) Namespaced names for operations.
Scribe: Straw poll: 2) Go back and revisit overloading
... Straw poll: 3) Status quo
<jeffsch> Agree that if we reintroduce overloading, we will need to do some careful (perhaps significant) work to get it right.
<Marsh> 2) Revisit overloading.
... 3) Status quo
ST: From GGF perspective, most critical thing is a binary decision between the new proposal from the F22F or overloading.
... Second is a binary decision on whether to go back to status quo.
... GGF would want us to at least do what's in the QName proposal.
JM: OK, we'll first adopt or not adopt changing status quo to deal with some degree of "overloading".
<Marsh> Revised straw poll: Accept Namespaced names for operations (or not)
Glen: Question: Is there an example of actual like syntax?
Gudge: Nope, no syntax changes.
JM: Asks "Does anyone object to adopting the proposal as worked out @ F2F and written up by Gudge"
Scribe: No objections.
JM: New question: Do we want to revisit overloading?
<dietmar> I'm muted, sorry
<jjm> LOL
<Marsh> 13 nos, 5 yes, 7 abstains.
JM: Are there any objections to not re-opening the overloading decision?
<jeffsch> ACTION: editors to fold in current proposal into main draft
JM: Do we want to put much more work into the HTTP binding or just let it be
<Marsh> JacekK will not :-)
... JJM: Wonders what the TAG's reaction will be? Recalls XMLP runaround
<JacekK> JM: it seems we should make the HTTP binding REST-friendly
... JS: my previous post received amazing lack of interest
<Marsh> http://lists.w3.org/Archives/Public/www-ws-desc/2002Jun/0102.html
JS: Wrote a proposal for how to encode many types into a URL. Paul Prescod said not needed to do that.
<Marsh> http://lists.w3.org/Archives/Public/www-ws-desc/2002Jun/0102.html
... 6a. Define encoding of complex types in a request URL
<jeffsch> http://lists.w3.org/Archives/Public/www-ws-desc/2002Apr/0183.html
JS: Issue is really about what to do with complex typed arguments.
... Made a proposal back in Apr '02 to define a mapping of complex typed data into a URI, but REST guy (Paul PRescod) says no no that's not the RESTful way of doing URIs.
<Philippe> David's proposal: http://www.w3.org/2001/tag/doc/ws-uri.html
Jack: Supports Jeff's proposal: close the issues w/o any extensions (only supports simple types)
Scribe: Arthur Ryman (AR): We may want to make that explicit in the HTTP binding.
JS: Agrees it'll be good to indicate what aspects of the abstract part of WSDL the normative binding will support.
Sanjiva: DOn't we need to figure out what to do with <message> before closign this?
JS: We should make decisions with the status quo (<message> exists) and then make changes later if needed.
... Are there any objections to closing issue 6a and 41 with no changes to the spec?
JM: Repeats the question
JJM: Does this mean the HTTP bindign will NOT be able to support complex types or that we'll have to use a differnet method for thta case (POST vs GET)?
JM: Thinks the answer is the latter.
AR: Agrees.
JM: No objections recorded: Mark those issues as closed.
<jeffsch> ACTION: Close issue 6a and 41
Scribe: ACTION: Editors to update the documents as needed and improve wording if necessary.
... JS/JM: Some discussion about I18N of URIs .. recommends we copy the right standard text from some I18N docs.
JM: "Doesn't matter what the solution is; just put a solution"
JS: Proposes editors just do that.
JM: No objections.
<jeffsch> ACTINO: Editors to work with jmarsh to include boilerplate for I18N
... ACTION: Editors to work with jmarsh to include boilerplate for I18N
... ACTION: Editors close 6b and 6d
Scribe: Talking about more HTTP binding issues.
... ACTION: Arthur to propose imprvements to the URL replacement mechanism.
JM: We'll keep issue 6e open pending Arthur's proposal.
Scribe: Now discussing issues 53 & 54
AR: Proposes we combine the two HTTP bindings into one that specifies the operation on a per-method basis.
... Will make a proposal on how to do that .. Phillippe will help.
Scribe: ACTION: Philippe: Re-do HTTP bindings to allow per-method operation names etc.
... AR will help PLH
... Now discussing issues 55 & 56
... Sanjiva states 55 is related to teh context proposal.
... JJM says its related to 14 also
CORRECTION: Related to agenda item 14 ..
JM: Wonders whether the message/part concept is indeed a fundamental thing or just something people have gotten used to because of WSDL 1.1
Umit: Many tools support taking existing stuff as Web services. In those scenarios you have to think whether hte asbtractions help or hinder people.
Jack: Thinks that if people think of more than XML protocols, then the message/part concept is natural.
Umit: Agrees;
AL: Disagrees completely. From a messaging guy's perspective a message is a message is a message; don't go looking inside that sucker.
Umit: Agrees with Amy. Thinks that both scenarios exist .. not one size fits all. (Missed stmt about <something> woman.)
AL: Agrees that the part concept doesn't disable the style of programming/networking that he's accustomed to.
... But it may make it less obvious to map.
<alewis> s/he's/she's/ (thank you very much)
Jack: Amy said that in messaging inside messages and hence parts are unnatural. However, in WSDL we have to do that so that parts of the message can be mapped to soap headers, http headers, url params etc..
... Doesn't think we need all XSD complexity to support the part concept.
JM: Outstanding actions for Roberto and Gudge to show example bindings.
<Philippe> http://www.w3.org/2003/02/13-ws-desc-irc.txt actually