OASIS and the Future of SAX

February 16, 2000

Leigh Dodds

OASIS is now the official host of the XML-DEV mailing list. The move has prompted some discussion concerning the role OASIS might serve within the XML community. The focus is on the future development and potential standardization of SAX. SAX (Simple API for XML) was developed collaboratively on the XML-DEV list and coordinated by David Megginson.

Jon Bosak began the debate, commenting that an OASIS technical committee should take over development of SAX:

I don't think that anyone owns SAX, OASIS included, but I strongly believe that further work on it should be conducted in a properly constituted OASIS technical committee so that the evolution of the specification from this point on is guaranteed to take place under a democratic process that is open to all interested parties, provides an IPR policy based on an open-source model, and is visible to the world at large. The xml-dev list supplies visibility but not the rest of what's needed.

Reactions to this suggestion were mixed. Peter Murray-Rust backed the idea, but qualified his support:

My argument was that SAX (V1.0) is now essentially static and as such is worthy of a formal home and curation (e.g., conformance testing as NIST/OASIS have done). SAX2 is dynamic, a product of David [Megginson] and this list and I would not support its transfer until it had passed through the same phases as SAX 1.0.

Other contributors were less positive about Bosak's suggestion. David Megginson believed that forming a committee would actually hinder development of SAX2 at its current stage, and disagreed that a committee would be a democratic replacement:

...it's democratic only to the point that the two, or three, or ten committee members get to vote, or perhaps to the point that the W3C or OASIS members get to vote, but not in the way that most of you might think.

He also expressed his desire to step down as SAX maintainer, and urged developers to begin thinking about the future of SAX. Simon St. Laurent was "violently opposed" to the idea of OASIS taking over development of SAX:

I'm disappointed to see it suggested that the most successful open-discussion based XML spec be taken to a more formal and less open process.

Bosak refuted the suggestion that OASIS represented a closed process. He pointed out that all technical discussions were publicly accessible. Membership is only required to actually contribute to the OASIS mailing lists (although this still presents a barrier to entry when compared to the accessibility of XML-DEV). On the point of democracy, Bosak observed that the development of SAX itself was not democratic:

It puts all decision-making authority in the hands of a single person. The fact that the person in question happens to be above reproach does not change the fact that this is a benevolent dictatorship. I don't buy the benevolent dictator model of standards development no matter how highly I regard the dictator of the moment.

He also claimed that ultimately SAX would be defined by how it was implemented by the major tool vendors, and not through the efforts of XML-DEV. His assertion was that OASIS is in a better position to enforce SAX as a standard, thereby properly safe-guarding the effort invested in it. Again, Simon St. Laurent disagreed:

I think you've somehow decided that a committee of paying members is a democracy while a well-coordinated mailing list project is a dictatorship. Simultaneously, you've placed the interests of large agencies and corporations above the needs of smaller organizations for small, flexible, and low-cost standards that work today.

Peter Murray-Rust mediated, pointing out that the backing of a standards body might encourage late-adopters:

[SAX has reached the medium-adopter stage and I am publicly extremely grateful to all the IT vendors who have adopted SAX. Without them it could have been very difficult to move further.] However we are now moving to late-adopters. Some of them may, by default, be happy because SAX is a component of any system they buy, but others might wish to be reassured that it has been implemented consistently in any product.

It's obvious that OASIS currently has an image problem in the XML-DEV community. Don Park, not impressed by OASIS, thought that it was time for more lightweight standards bodies:

What we need are smaller standards organizations, each with a well-defined and focused charter. These organizations should do no more than approve or disapprove standard proposals, no matter who created them....

Whether OASIS can improve its image remains to be seen. Ironically, the debate was cut short earlier in the week by technical problems at OASIS that made XML-DEV unavailable. These problems couldn't have come at a more inopportune time. What will happen next is uncertain, but David Megginson's expressed desire to give up the coordination role means that change in the way SAX is run will be inevitable.

The complexity of groves was a topic discussed in an earlier edition of XML-Deviant. In a spin-off from the original discussion, Len Bullard challenged W. Eliot Kimber to produce a description of groves in 50 words. Kimber accepted the challenge:

Groves, by providing a generic, basic, universal abstract data model, provide a formal basis for defining data models for specific data types, e.g., XML, VRML, relational tables, etc. This provides a basis for standardizing abstract data models and enables the application of generic processing to data in a provable, testable, way.

Didier Martin, in his inimitable style, suggested to Bullard that it was time to "get practical," and work through a concrete example using VRML:

The best way to know Groves is to use them practically. So, what is the problem in 3D land that we can try to resolve with Groves? By discussing the model or using Groves to model it, we will know the difference between Groves and the monster of the Loch Ness.

Bullard agreed, and posted a brief overview of VRML as reference material for the list. Advocating using XML-DEV to explore the issues, he commented further:

The problem first is for us to prove to ourselves and the rest of the list the utility [of Groves]. The way is to use them and apply them here in examples until we are all conversant. Until there is an AHA from the rest of the markup community, then our biggest problem is the obscurity of the descriptions.

It will be interesting to see this discussion play out, as it could provide much needed introductory material on groves. It's worth mentioning that groves were being discussed on XML-DEV as far back as three years ago without significant results. Perhaps this will be the start of something more concrete.

The Gutenberg Project is entering into a joint initiative with the HTML Writers Guild. The purpose of the initiative is to produce XML versions of the 10,000 volumes available in the project's archives. Frank Boumphrey pre-announced the project this week, seeking volunteers to contribute to the design of suitable DTDs.

This is a worthy project, and deserving of some attention. If nothing else, it's likely to be an excellent source of data for XHTML and XSL developers. In addition, it should serve as an interesting experiment in collaborative markup and produce an impressive electronic archive.