RE: [Bug 4632] Use of IRIs

Thanks for sharing your thought on this issue. I assume that you agree with points 1, 2 & 5. If so, thank you for your support on that.

#3
I would really like to avoid going down the path of debating which version of which implementation has what level of support because we may get lost in the details. The important point to consider here is that internationalization of sml:uri is already possible today without making any change to the spec or to the existing implementations. Requiring implementations that use an XML 1.0 processor that currently does not support RFC 3986/3987 to support it places an undue burden on those implementations. The xml processor in .net framework and the native code implementation (msxml) does not support RFC 3986/3987. Even if the support gets added today, it will take many months before it is adopted for enterprise critical apps due to LOB app compatibility issue and other business constraints.

We have spent a lot of time debating this issue. IMO, it would be more productive to focus on issues with intrinsic value to SML, such as the excellent proposal from IBM on SML references.

Re. comments on #4:
The assumption is that SML processor supporting schema 1.1 will be built on top of an XML schema 1.1 processor (rather than from scratch) and thus it can relegate the responsibility of supporting IRI RFCs on the underlying processor.

The point about the possible Interop issue is an interesting one. I am sure we can find more such cases (not just those related to anyURI) where the 1.1 behavior is different from 1.0 behavior. We will probably need to come up with a conflict resolution policy that describes how such cases will be handled by the future versions of SML. We should discuss them when we discuss aligning the next version of SML with XML schema 1.1.


From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On Behalf Of John Arwe
Sent: Monday, September 17, 2007 9:26 PM
To: public-sml@w3.org
Subject: Re: [Bug 4632] Use of IRIs


> 3.    Most XML schema 1.0 processor implementations do not support IRIs. This
> means that SML implementations that use an off-the-shelf schema 1.0 processor
> will need to provide additional implementation to support IRIs. Implementing
> RFC 3987 (46 pages long) is non-trivial amount of work. This is an unnecessary
> implementation burden given that internationalization is already possible as
> described in #1.

Excellent, this will make a decision much simpler. Please share the research you did that led to this conclusion: which implementations you surveyed, which support IRIs, and pointers to the relevant reference(s).

> 4.    If we decide to align SML spec with XML schema 1.1 in a future release,
> we will automatically get IRI functionality as schema 1.1 is aligned with the
> IRI RFCs.

This does not seem to follow from the rest of your email, especially in light of MSM's addition.  If the value space of XML Schema 1.0 anyURI is a superset of the IRI value space, what exactly did you gain by switching schema processors except the unchecked assurance that the value can play the role of a URI (1.0) or IRI (1.1)?  It seems like the SML processor supporting XSDL 1.1 would still have to change to implement those 47 long pages of IRI RFC, no?

It would also seem to introduce new ways to break interop, although since I have yet to read XSDL 1.1 I'll rely on the schema experts to correct me.  If I have a schema document valid using both XSDL 1.0 and XSDL 1.1, a bi-lingual producer told via run-time option to produce an SMLIF instance as XSDL 1.1-compliant, and a consumer told to interpret that instance as XSDL 1.0-compliant, I seem to have just changed the interpretation of the anyURI values w/o touching the document.  And unless said values are actually de-referenced (vs only being compared as document aliases) it seems unlikely that this will be detected...assuming the coders were lazy and/or did not realize that "anyURI" validation did not actually restrict the value space to RFC3986-acceptable characters.

Best Regards, John

Street address: 2455 South Road, Poughkeepsie, NY USA 12601
Voice: 1+845-435-9470      Fax: 1+845-432-9787

Received on Wednesday, 19 September 2007 08:50:01 UTC