[relax-ng] Minutes for RELAX NG TC Telecon 2002-07-18

[Michael Fitzgerald <mike@wyeast.net>]

Minutes for a RELAX NG TC teleconference held 18 July 2002 at 9:00 am EDT
(UTC -04:00)

Attending

James
John
Kohsuke
Mike
Norm

Not Attending

David
Josh
Makoto

Agenda

1. Annotations

John: I think annotations inhibit the freedom of compact syntax translators
James: What about the reasons I gave? [in an email message]
John: I'll look it up
James: What do other people feel?
Mike: I think the current syntax is fine and that we need them
John: I see a couple of cases here, one for default attributes, another for
generating code, the RelaxNGCC case
James: No this case is different, RelaxNGCC is a parser generator
Kohsuke: Like yacc or castor
John: Then the adjustments would be small
James: There are two kinds of translations (1) just getting the same class
of document, essentially a validator, and (2) one for human consumption.
Trang is very constrained.
John: we need a double-barreled conformance statement
Kohsuke: we could drop following annotations as a compromise
James: yes, elements are harder
Norm: I want to keep them, we want to be able to move them back and forth
easily
John: but they won't be roundtripable
James: the output won't be identical, but so what?
James: I think annotations are essential, useful for a hundred little things
John: what if we did attribute annotations only?
James: leaves out a lot of things...strongly feel that we need them
John: I am willing to bow to the consensus
Kohsuke: I don't feel strongly that we need to drop element [following]
annotations

[No vote taken, not really necessary]

2. Conformance

John: we need different levels of conformance for validators and translators
James: I don't see them as vastly different
John: a translator may throw out annotations, and just produced simplified
syntax, for example
James: I see three types: (1) validator, (2) a general translator that spits
out XML, (3) a "structured isomorphic" [John] translator that maintains
structure
Norm: I am having a hard time mentally with this...
Kohsuke: what is the difference?
James: (1) doesn't have to provide an XML serializer
Mike: we need a conformance statement for all three
John: four really, one for document conformance
James: I will take an action item...

ACTION: James will write a conformance statement for inclusion in the spec

3. Redundant trailing ,|&

John: [uncaptured statements in support]
Norm: I think we should avoid it
Kohsuke: [uncaptured clever workaround]
James: these will present an implementation complexity
John: I realize now that they will cause an implementation complexity [and
relent]

James: other items?
John: tutorial [John has already done a preliminary draft]

4. Tutorial

John: there were lots of subtle changes, such as in the namespace section
James: best if tutorials were all of one source, then transformed to
separate documents
Mike: I wonder if we could combine tutorials, all the same document
James: too complex, but do you support this as a committee document?
Mike: yes
Kohsuke: I could go both ways
John: the compact syntax is "pedagogically superior"
Norm: did you do your draft in docbook?
John: No, just HTML. Original in docbook?
Norm: Yes...I'll take an action item to put it into docbook, since I know a
little more about it than most on the committee...

ACTION: Norm will combine tutorials into on DocBook document

John: status of XML Schema output [of Trang]?
James: yes, an endless list of things to do...I am finished with the parts
one can do with DTDs...one of the hardest parts is representing
namespaces...have to use separate documents, approximate...two remaining
troubles are element types and ambiguous content models...I am looking at
the work of a Finnish research on ambiguous content models
John: can you release a beta?
James: I think I will call it alpha!
John: in all my spare time these days I am working on a compact syntax for
XML Schema

Kohsuke: Did the guy who wants to do RELAX NG with PHP get a hold of you?
James: Yes. I think he wanted me to write out all the [functions/algorithms]
to do so but I said no.

James