[relax-ng] Minutes for RELAX NG TC Meeting 2002-02-28

[Michael Fitzgerald <mike@wyeast.net>]

Minutes for the RELAX NG TC telcon held 28 February 2002 at 10:30 am ET
(UTC -05:00).

The next RELAX NG telcon will take place at 9:00 am ET (6:00 am PT) on
Thursday, 7 March 2002.

Attendees

David (late)
James
John Cowan (observer)
Kohsuke (late)
Makoto
Mike
Several observers from Vitria (didn't catch their names)

Not Attending

Fabio
Josh
Norm

Minutes

1. Erratum. End-tag typo in sec 3.1 of the spec: </a:document> should be
</a:documentation>. Discovered by Michael Smith. See
http://lists.oasis-open.org/archives/relax-ng/200112/msg00024.html.

Approved.

2. Erratum. Currently, foreign attributes on <param> element in RELAX NG
schema for RELAX NG are not permitted. James will add <ref
name="common-atts"/> to the <param> pattern. See
http://lists.oasis-open.org/archives/relax-ng/200112/msg00044.html.

Approved.

3. Erratum. Example in tutorial violates restriction on <interleave>,
producing the error "overlapping
element names in operands of 'interleave'" in Jing or "names of elements in
<interleave> are competing each other" in MSV. See
http://lists.oasis-open.org/archives/relax-ng/200202/msg00007.html.

James will post a corrected schema to the list for approval.

4. Erratum? DTD Compatibility prohibits choice name classes with ID/IDREF.
John Cowan had suggested this. See
http://lists.oasis-open.org/archives/relax-ng-comment/200202/msg00008.html.
He wants

element foo | bar | baz {
	attribute id {xsd:ID},
	text
	}

to be allowed in the non-XML syntax, that is >= one name.

Approved. We will allow more than one name and add this as an erratum for
the Compatibility spec.

5. Future program of work.

-Non-XML syntax?

Approved. The committee will produce a specification for a non-XML syntax
for RELAX NG.

-Key/keyref?

James: key/keyref is a missing piece in RELAX NG. We need a streaming
solution. What we discussed before 1.0 was a bit of a hack. We need a fully
baked proposal, coming as a spark of inspiration from someone on the list.
David: We should build a list of use cases, then figure out what we want to
do.
Mike: It seems important enough that we do it right.
John: Important but not imperative.
James: RELAX NG will survive without it but...

-Associating schema with instances?

Committee will not produce a solution, but will publish suggestions in a
FAQ. We will look elsewhere for a general solution. Sean McGrath's XPipe was
suggested by John. See http://xpipe.sourceforge.net/.

-Test suite?

James: we just have an informal one now
Makoto: I think it would be good to have an official test suite
James: I am not sure we would each individually approve thousands of cases
David: Couldn't we ask OASIS to take this on?
James: I have already done a lot of the donkey work
John: Donkey work does not make one a donkey
James: NIST finds every assertion in a spec and creates a test for it
Mike: Perhaps Josh would have some insight here

-FAQ?

Yes we need to finish this up.
James: Trouble is figuring out who will do what...

-New RELAX NG version?

James: We could do something about min/max (minOccurs/maxOccurs
functionality)
John: You could just do this with preprocessing

No we will not consider a new version of RELAX NG quite yet.

-A RELAX NG datatype library?

James: most folks are using the XML Schema datatype library...
Makoto: XML Schema datatypes are too complicated and too buggy
David: we should do something about regular expressions, but perhaps hide
the [spaghetti] from users
John: I'd like to see a fine-grained pull-in [include] mechanism for RE
processing
Makoto: Every implementation of XML Schema datatypes omits some features
James: Instead of character to character comparison we could use something
like <list> [token pattern matching]

Next week's meeting will be 1 and 1/2 hours earlier due to Kohsuke's
availability at that hour.