IRI, Directionality and Ruby in ITS

Hello Martin, Richard, all,

This is related to the ITS issues

http://www.w3.org/Bugs/Public/show_bug.cgi?id=3466 ("Existing ruby markup")
http://www.w3.org/Bugs/Public/show_bug.cgi?id=3494 ("termInfoRefPointer
referant's data")

These two issues have in common that they have relations to external
specifications:
1) Ruby TR for 3466
2) Directionality as defined in XHTML 2.0 (if it a REC in time,
otherwise HTML 4.01) for 3466
3) The IRI spec for 3466

I am worried about these relations. ITS allows for selecting nodes via
global or local ITS markup, and attaching information to them, or
pointing to existing information. What happens after selection, adding /
pointing, is not the business of ITS anymore.

That is, an implementation of ITS processing expectations, see
http://www.w3.org/TR/2006/WD-its-20060518/#conformance-product-processing-expectations
does not process

1) Ruby information as demanded by the Ruby TR
2) Directionality information, relying on the BIDI algorithm and its
manipulation
3) IRIs

Of course, an implementation *on top of ITS* needs to do such processing
to be able to do s.t. useful. But since such processing happens after
ITS, I think the ITS spec cannot say anything normative about what
should happen. If we would say s.t. normative here, we would need to
provide tests about it. But I don't see how to test *with ITS alone* that

1) an ITS rule like <its:rubyRule select="//myns:ruby"
rubyPointer="self::*" rbPointer="myns:rb" rtPointer="myns:rt"/> implies
processing for the myns:ruby, myns:rb and myns:rt elements in the sense
of the ruby TR
2) an ITS rule like <its:dirRule select="//*[myns:dir='rtl'] dir="rtl"/>
 manipulates the Bidi algorithm via the value of the dir attribute
3) an ITS rule like <its:locInfoRule select="//someElem"
locInfoRefPointer="@myns:locInfoRef" points to an myns:locInfoRef
attribute which contains an IRI as a value.

So my proposals:
- delete conformance clause 2-2 in the draft (which is about normative
references to the ruby spec and XHTML 2),
- cite Ruby, XHTML 2.0 (or HTML 4.01) and the IRI spec only
informatively, and
- tell implementers in a non-normative section what they should do - on
top of ITS.

What do you think? Since both of you have been heavily involved in the
development of the three specs, your opinion would be of great value.


Regards, Felix