Minutes of Encoding Task Force, 19 February 2002.

Start time: 8am pacific, 11am eastern, 4pm UK, 5pm Europe
Duration: 60 minutes

(a) Agenda review and AOB

AOB - need a volunteer to report status at the F2F; decided to play by the
ear
AOB - Jacek requested for a copy of the minutes from the our last telecon;
chair to request Yves to post it on the server ASAP

(b) Chair and time for next call

Shall we meet again this week or in Cannes? Postponed as the last item

(c) Review actions items,

* All to investigate alternative array serializations for next ETF call
* HFN to ping Paul Cotton regarding his email on alternative array
serializations

DONE, sent via e-mail; I couldn't find any links

* HFN will propose changes to spec and primer to resolve ed issue #17

DONE, http://lists.w3.org/Archives/Public/xml-dist-app/2002Feb/0318.html

* MJH to remove "(possibly defaulted)" from rule 2

PENDING

* Asir to contact Martin Gudgin regarding need for removal of NOTATION
element. ETF suggests leaving it in and adding pointer to XML schema warning
about these types. Asir to draft warning text.

Partially completed - contacted Martin Gudgin. Postponed discussion

* HFN to check if issue 113 is closed
* HFN to draft resolution for 16

DONE, 113 and 16
http://lists.w3.org/Archives/Public/xml-dist-app/2002Feb/0231.html

* JK to check that RPC side of issue 78 is closed and to propose rewrite of
root attribute section to improve clarity.

DONE
http://lists.w3.org/Archives/Public/xml-dist-app/2002Feb/0301.html

Postponed discussion

* MJH to ping TBTF to check if issue 44 is a TBTF issue

DONE, by Henrik and it is indeed a TBTF issue

* Issue 29 is ready to close. Yeah, but what is the action item?

This is a long outstanding one. We might have an new issue. We made gross
changes to encoding - href. NEW ACTION: Jack to send a reminder note to
David to close issue #29 through the XML Coordination Group and inform him
about changes to encoding, href

(d) SOAP NOTATION element decl - if Asir provides any new info from Martin

We decided to forward our decision, (leave the legacy types, change Notation
attribute decl to element decl, add a reference to XML Schema warning about
these types), to the working group. We reviewed the issue outlined by Asir
in,

http://lists.w3.org/Archives/Public/xml-dist-app/2002Feb/0226.html

ACTION ITEM - Asir, contact Yves to open an issue and send a proposed
resolution text to the dist-app

(e) Issue 180 - Parameter ordering in Section 4.1

Jacek summarized this issue. There were some discussions on parameter
ordering on dist app,

http://lists.w3.org/Archives/Public/xml-dist-app/2002Feb/0161.html

This thread outlined several issues and options,

- don't say anything about ordering
- don't say anything about return value accessor
- don't say anything about method signature, yeh WSDL
- just say that order is significant

We discussed and considered three options,

- remove ordering ('cos struct has no order)
- make it consistent (out and in/out parameter should be in an same order,
if method signature available (meta data))
- say nothing about ordering, what method signature (out of scope) .. say
nothing about it. Ordering can be specified in a description

Resolved to say nothing about parameter order and nothing about method
signature.

NEW ACTION: Henrik to draft the proposed resolution for issue 180


(f) Multistructs - I didn't see any hyperlinks to discussion

Asir reported that there weren't any direct discussion on this topic in
dist-app. Jacek raised this a couple of times as part of other issues. Jacek
described what is a multistruct. It is the generic compound type, something
weaker than struct. Why is it an issue? No one uses it and its
implementation complexity. struct and array can be represented using
multistruct; it is a pain to support them.

Asir pointed out that RDF folks will need generic compound type to represent
bags and sequences. These containers permit duplicate values.

Jacek said that SOAP data model is a generalization of  the common features
found in type systems in programming languages. Henrik said that it is
difficult to come up with a model for only programming languages.

Asir pointed out that SOAP data model is a generalization of the common
features found in type systems in programming languages, databases and **
semi-structured data **. And, webMethods currently uses generic compound
type of the SOAP data model and did not report any implementation
complexity.

We decided to take this further via e-mail.

(g) Other issues

Issue 78

Many expressed concerns about removing root attribute

Considered other options,

- Root is the first element in the body or labeled root
- In many cases, instance data is self explanatory, for root; if it is
ambiguous then use root. This opens up a possibility to leave it ambiguous
- must mark known roots explicitly
- must mark non-roots explicitly, ease of moving forward

Misc

- Elminate current obscure wording
- observed that multi-refs serialized inline poses no problem
- if you mark the root, then it is less data
- yeah, in practice, if you mark the non-roots, then it is less data

Jacek's e-mail also proposed to change the name of the root attribute. 'cos,
the current naming is inconsistent.

Resolved - must mark non-roots explicitly, to use the rewrite from Marc;
majority to leave the current naming inconsistency as it is and raise a
separate issue.

New Action: Jacek to draft a proposed resolution for issue 78

Miscellaneous items

New Action: ETF, discuss alternate array serializations via e-mail

New Action: Asir, send a list of ETF issue resolutions to David Fallside


Time for next call - decided to play by the ear based on F2F outcome.



New and Pending Action Items

(a) Jack to send a reminder note to David to close issue #29 through the XML
Coordination Group and inform him about changes to encoding, href
(b) Asir, contact Yves to open an issue on XML Legacy Types in SOAP Encoding
and send a proposed resolution text to the dist-app
(c) Henrik to draft the proposed resolution for issue 180
(d) Jacek to draft a proposed resolution for issue 78
(e) ETF, discuss alternate array serializations via e-mail
(f) Asir, send a list of ETF issue resolutions to David Fallside
(h) MJH to remove "(possibly defaulted)" from rule 2