XMLP WG Telcon minutes, 21 January 2004
1. Roll
Present 17/12
BEA Systems, Mark Nottingham
BEA Systems, David Orchard (scribe)
Canon, Herve Ruellan
IBM, David Fallside
IBM, John Ibbotson
IONA Technologies, Seumas Soltysik
Microsoft Corporation, Martin Gudgin
Nokia, Michael Mahan
Oracle, Anish Karmarkar
SAP AG, Gerd Hoelzing
SAP AG, Volker Wiechers
SeeBeyond, Pete Wenzel
Sun Microsystems, Tony Graham
Sun Microsystems, Marc Hadley
Systinet (IDOOX), Jacek Kopecky
W3C, Yves Lafon
W3C, Carine Bournez
Excused
Canon, Jean-Jacques Moreau
IBM, Noah Mendelsohn
IONA Technologies, Mike Greenberg
Microsoft Corporation, Jeff Schlimmer
Oracle, Jeff Mischkinsky
Systinet (IDOOX), Miroslav Simek
Regrets
DaimlerChrysler R. & Tech, Mario Jeckle
Absent
DaimlerChrysler R. & Tech, Andreas Riegg
2. Postponed
3. Approval of 14 January 2004 telcon minutes: approved without objection
4. Review action items
No one in WG disagreed with status of actions as listed in the agenda
5. Status reports
- Media types registration. MarkN reported no change in progress.
- Anish reported no change in XMLP/WSDL taskforce. There has been no meeting of
the task force. One proposal has been made, but it has received no responses.
6. Attachments
Review of MTOM and XOP documents in preparation for WD publication
MAJOR ISSUES
Lossy:
MarcN: XOP is lossy
MarcH: The only type info that is preserved is attached to stuff
that is optimized. That type info can be used by bindings in an
intermediary to preserve optimizations.
MarcN: talking about data model, data model for optimzation may be different
than dm for app. The relationship between 2 is magic. We can't say that app
dm is preserved.
DF: We are agreed on the nature of the model, if we see anything contrary then
call it out.
MN: Did we resolve DM in data model has to be base64?
DF: Let us continue into docs.
Parts:
Issue 440 was closed without objection with the consensus that was reached in
the email thread starting at
http://lists.w3.org/Archives/Public/xml-dist-app/2004Jan/0054.html.
More specifically,
- XOP can have multiple inclusions of the same binary part
- the HTTP binding can have only one inclusion of any binary part
- HTTP binding/MTOM and XOP allow unreferenced attachments but they are outside
the SOAP processing model.
Several WG members preferred MarcH's second wording proposal for describing that
unreferenced parts are allowed.
Base64:
Jack: Abstract feature might not be base64.
MN: Locking down base64 in XOP seems severe, lockdown in HTTP binding is OK.
Gudge: If we allow more than base64, these need to be defined or extensibility
provided for.
DF: Compromise suggestion: For the purposes of publishing a WD, we will permit
only base64 at all levels, i.e. xop/mtom/binding, and we will ask for priority
feedback on that decision in the document.
WG members agree to proposal without objection.
Remove Abstract Feature out of MTOM
DF: We decided at f2f to keep abstract feature, and we cannot re-open such
decisions without new information.
DO: The reduced siz of the MTOM doc is a reason to reconsider
Gudge: All the words in MTOM sec 2 are elsewhere in XOP, and so I prefer
to remove the feature.
Some WG members say they wish to keep the abstract feature.
The Chair determines that no-one is very strongly in favour of removing it.
DF: We will keep the feture in this WD version, and ask for feedback.
Xbinc:Include in content (issue 446)
DF: Noah effectively resolves 446 with the words he proposes for XOP section 3.
Specifically, these elements are not allowed in source data model.
DO: escaping is one solution
Gudge: prefix remapping is a possibility
Issue 446 was closed with Noah's text. There were no objections.
Review of comments and proposed text, per agenda.
Lots of specific discussions on what Herve is supposed to do as editor.
DF: We will publish a draft of the WD to the WG, on January 26/27, allow a day
or two for review, and then decide whether or not to seek WD publication at
next week's XMLP WG telcon (January 28).