See also: IRC log
<alewis> scribe: alewis
minutes accepted.
Review of Action items [.1].
[Interop]
? 2006-11-30: [interop] John Kaputin to create a test case
with "required=false".
? 2006-12-14: [interop] Jonathan to fix transferCodings -
add control group
[WG]
? 2006-09-21: Jonathan to check periodically that SPARQL has
added schemaLocation.
? 2006-12-14: plh to come up with a more detailed proposal for
CR112 if possible
? 2007-01-04: Paul to report back on which test cases in the
WSDL test suite fail the basic patterns, with
suggestions on how to address the issues.
? 2007-01-11: Jean-Jacques to provide more analysis on how
difficult it would be deal with a Policy that
only contains an MTOM policy assertion
DONE [.3] 2007-02-01: Arthur to review the WSDL 1.1 identifier spec.
DONE [.4] 2007-02-01: Jonathan to provide an enhanced option 1 for issue
117 by next week, using bits from the other
proposals as indicated above.
Current Editorial Action Items
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/2007Feb/0002.html
[.4] http://lists.w3.org/Archives/Public/www-ws-desc/2007Feb/0039.html
jonathan: any objections to forwarding comments?
<scribe> ACTION: jonathan to forward comments on one-way mep to XMLP working group [recorded in http://www.w3.org/2007/02/08-ws-desc-minutes.html#action01]
jonathan: don't understand why this exists; it doesn't work properly
arthur: exists to match wsdl 2.0 component designators
jonathan: has big problem with imports and includes
jacek: component identifiers do work properly for wsdl 2.0.
(or maybe jacek was talking about the 1.1 identifiers?)
arthur and jacek seem to think that the identifiers are okay; jonathan now uncertain of his analysis.
jacek: since imported things are in their own namespace, everything actually does work.
jonathan: will take another look, trying to find a case that's unclear as an example.
arthur: targetNamespace isn't required in 1.1, is it?
defer for a week, review again and discuss; jonathan continues to believe that there may be a simpler approach.
jonathan: call for volunteers to review.
jonathan has tried to write up option one in more detail.
needs massaging, to incorporate clarifications.
youenn has clarified option two, as well, making one and two fairly compatible (?)
discussion of parameter encoding, issues raised by jacek in email.
youenn: need to remove ampersand, colon, equals (all special URL characters that might have a special use in a URL) should be encoded, to be safe.
youenn offers to send a list of characters which may have to be encoded.
discussion of which ones need it, attempting to talk out the list.
jonathan: if we update list to include query parameters, does that solve it?
jacek has pointed out that question mark and slash are in wrong place.
jonathan will move those two.
jonathan: okay, what do people prefer?
discussion of some other characters (tilde, exclamation point)
three changes: encode query parameter separator, move two characters, add provision to encode query parameter separator, and change tilde to exclamation point
ready to adopt?
no objections heard
RESOLUTION: CR117 resolved with option one as modified.
jonathan has had offline discussions with various folks on handling the http binding.
proposal is to change transfer-coding to content-encoding in the syntax, in the name of the property, and to clarify the description.
call for comments: none heard
RESOLUTION: CR 143 closed with the resolution: change transfer-encoding to content-encoding, clarify description.
jonathan can't remember why this wasn't closed with acceptance of the proposal from last week's minutes.
anything thrown out with ignoreUncited=true must be nillable; postponed because youenn wasn't around.
<Jonathan> http://lists.w3.org/Archives/Public/www-ws-desc/2007Feb/0009.html
jonathan: this is fairly esoteric
jacek, jonathan: should we be describing receiver behavior?
arthur: it's not about behavior, it's about the appearance of the message on the wire, and whether that can be unambiguous.
jonathan: is this really worth fixing?
jacek: interop problem appears when two different services are created from the same wsdl, and they differ in how they handle missing bits.
jonathan: best way to handle that is to tell people not to combine default and nillable.
arthur: add an assertion (a should)
youenn, jonathan: maybe just change "must" in proposal to "should" ?
jonathan: want this resolved. maybe no one really cares and it could be closed with no action?
arthur: prefer to err on the side of interop rather than flexibility.
jonathan: reiterate proposal in email, with grammatical cleanup and the additional should.
RESOLUTION: CR 146 closed with the email proposal from jonathan, as modified.
the safety flag as semantics.
<pauld> scribe: pauld
jacek: introduces the issue, WSDL
doesn't describe semantics, apart from "safety" sawsdl is a
framework for annotating WSDL with semantics
... as chair of the SAWSDL WG i believe our specification is a
better mechanism than the WSDL WG defined annotation
arthur: I don't agree it's semantic, it allows you to understand when you can use GET
marsh: I thought the use-case was to flag POST
jacek: and you can't explicitly say an operation is NOT safe
marsh: if an operation is known
not to be safe, then you can't bind to GET
... it's a syntax discussion, and the impact on our component
model
... we should be aware it could be hard to use and changes our
syntax
... my personal position is it is a semantic annotation and
cleaner in an extension, and the GET is a small feature
... we may receive comments from the TAG if we removed our
safety feature
<alewis> scribe: alewis
paul: it seems clearer if it is left to the sawsdl working group
jonathan: can we leave this to
another working group at this stage?
... an alternative is to have sawsdl define on top of the
existing safety property, so there would be two syntaxes for
the same property.
paul: would be nice to remove, though, so that there's no possible confusion, and it can't get lost.
jonathan: tag likes safety to be visible and easy to use; maybe better to do it as one of several semantic annotations, but might be late to do it.
jacek: proposes actually asking the tag whether this could be moved to sawsdl, and to delay resolution until they respond.
jonathan, jacek: note that this is not dropping something that is not at risk, but moving it from one CR to another CR.
<scribe> ACTION: Jacek to approach TAG on subject of moving "safety" annotation from WSDL specification to SAWSDL [recorded in http://www.w3.org/2007/02/08-ws-desc-minutes.html#action02]
<pauld> presumably the SAWSDL URI could have value outside of WSDL, so the TAG might prefer such a solution
instead of generic "message label", replace with "in" and "out".
<asir> +1 to close without any action
RESOLUTION: CR 129 closed with no action (based on misinterpretation of message labels)
<Jonathan> http://lists.w3.org/Archives/Public/www-ws-desc/2007Jan/0000.html
jonathan posts url of mail from arthur containing a shorter formulation
RESOLUTION: CR 132 closed with the first two assertions from the original mail, last two from arthur's amendment.
RESOLUTION: CR 136, 137, 138 resolved as editorial
jacek: wait, isn't this about the property, so it *isn't* actually a union? property always populated with specifics.
question of whether #any or #none make sense. argument is that it doesn't make sense, since there's no point in defining a fault if you don't actually define it.
jacek: but #other still makes sense. and therefore we need to change the content model, modeling it on interface message reference.
replace current with #other or #element.
general agreement that #any and #none actually do make sense in this context.
arthur: in that case, it's exactly parallel to message?
RESOLUTION: CR 138 to be resolved by making faults identical to interface message references.
jacek: procedural not used
elsewhere, document-oriented not used elsewhere.
... network service also only used in part two abstract.
... modify abstract in part two to match part one. wsdl is a
language for describing web services.
RESOLUTION: CR 139 closed as editorial, editors instructed to make part two use same definition as part one.
<Arthur> Concerning the previous issue on the Abstract, we should refer to http://www.w3.org/TR/ws-arch/ for the definition of Web Service
<Arthur> i.e. CR139
jonathan: there's a header that canon and wsos2 are both using; just add it, since it aids interop?
jacek: no, this is an entity header; can't be used in this fashion.
<Arthur> CR 148 I think Amy
<scribe> ACTION: jacek to investigate soap and http with respect to soap action header [recorded in http://www.w3.org/2007/02/08-ws-desc-minutes.html#action03]