Birth of a Community

February 9, 2000

Leigh Dodds

This week the XML-DEV mailing list moved to its new home with OASIS. For the past three years the mailing list has been hosted by Imperial College, London. XML-Deviant talked to Professor Peter Murray-Rust, one of the founders of XML-DEV, about the role that XML-DEV has played in the XML community.

Making a List

XML-DEV was born on February 21st, 1997, when Peter Murray-Rust welcomed the list members, and thanked Henry Rzepa for taking on its setup and maintenance. Jon Bosak, chief instigator of the original XML activity, expressed high hopes for the list:

This list is certain to become a major resource as we begin to develop applications of XML.

Bosak's comment has since been vindicated. XML-DEV now has nearly 1600 subscribers and provides a content rich forum for XML developers around the world.

The early list archives make for interesting reading. There is a sense of pioneering spirit among the small group of developers working together to implement a fledgling standard. It's also clear how much effort Peter Murray-Rust put into the list: as a frequent contributor he guided many of the early discussions. I asked him about the aims behind the list's creation:

I was part of the initial XML working group (later the SIG) and passionately concerned that XML succeed. Henry and I had a hidden agenda, which we can now reveal—we needed XML for interoperability in chemistry. I had developed Chemical Markup Language (CML) in SGML and had tried to "sell" it to chemistry—not a chance. SGML was far too complex, and the practitioners didn't always see the problems it caused normal mortals. Our "looking glass war" was therefore to put our efforts into helping XML succeed—globally—and then come back to chemistry with the major battle won. Therefore XML had to be universally adopted and easy. If we could add 1% help to the XML effort it was worth it.

I had seen too many information "standards" fail because they were simply too complex to implement. It is probably 10-100 times easier to write a standard than to implement it (we see that with Namespaces!). I wanted to make sure that XML always had an implementation as it emerged, so that the drafters could see the problems incurred as they arose. [I contributed to that ideal by developing the JUMBO browser and presenting it to the community.]

Three years on, the W3C has moved to formalize just this process by introducing the Candidate Recommendation (CR) phase for specifications. CRs are provided to allow time for implementation of a specification. Murray-Rust expanded on the importance of implementations and an open approach:

I also believe strongly in Open Source as a powerful tool in standards development. An open approach winkles out the hidden semantics, which are often not documented and are implicit in a single vendor's approach. I think initially some people saw XML software as having a significant market potential, continuing from the SGML tradition of overpriced non-interoperable tools. The reality was, of course, that software modules are becoming increasingly cheap. So one great initial achievement of XML-DEV was to foster the Open Source idea. I remember one company ... who initially offered their software on a restrictive license. Several postings from the list and later they announced, "We listened!" and revised the license.

Initially, also, there were no public fora for XML. The editorial group was about 12 with a wider working group of around 100 experts. None of this discussion was public. One great achievement of XML-DEV has been to give an opportunity for meritocracy. Many people have emerged through this list and some have been invited to be experts in the closed W3C activities. Another important role is that the list could affect both decisions and practices of the W3C groups—not formally, but through the voice of a respected community. I was always very careful to make our non-statutory role clear, but there were occasions when the W3C specifically communicated with XML-DEV.

Successful Elements

The list has had a number of successes over the years, not the least of which is its role in helping to popularize XML:

Although it's not obvious now, XML was a Cinderella when it started—it was not on the W3C's home page, but buried as "SGML activities" or similar. It could easily have remained hidden and died. So there was a need for publicity, demonstrators, etc. The major credit for publicizing and promoting XML is, of course, Jon Bosak's, but I believe XML-DEV helped by giving a sense of community and providing some ammunition.

Probably the most trumpeted success story is that of the SAX API. SAX is a de facto standard that is now supported in many XML parsers. It was produced as a collaborative effort through XML-DEV:

The XML activity made it very clear that the recommendation should not contain any implementation details, but there was a clear need. I catalyzed the list into action and after 2 early attempts, David Megginson "conducted" us through SAX. I take great pride in seeing that almost every new software tool includes SAX.

The history of SAX has already been documented by David Megginson, but it is worth mentioning that the origins of a simple API date back to the early days of XML-DEV. (A point of historical interest: the acronym JAX was rejected as it was noted that "jax" is Irish slang for "toilet"!)

XML-DEV has been successful in other ways as well:

The list has also produced protocols that have informed the W3C process. XSchema and DDML have, I believe, been valuable to the Schema group. SML, though controversial, is an example of how new approaches need a place to be aired. And it is to the list's credit that a devolved organization has been developed.

The greatest achievement for me, however, is creating a global virtual community. These are events that happen—they cannot be planned or created. The combination of "building castles in the air, from air" (Brooks) with the ability to reach any part of the planet within seconds can be intoxicating. Merit—including commitment—and respect for it is what ultimately counts.

Agents Provocateurs

XML-DEV has a reputation for being an extremely vocal forum. Criticism for W3C specifications can at times become quite lengthy. However, these contributions are of very high quality, and certainly don't fall on deaf ears. Notably, at the end of last year XHTML was returned to Working Draft status after a discussion of the "Three Namespaces Problem" on XML-DEV.

Murray-Rust commented further on the list's role in analyzing the W3C specifications:

In the early days I was pro-active in steering the list to implement what the editors and the XML-SIG produced in their drafts. I knew that those groups listened to what we published. To have discussed more widely whether XML was "right or wrong" would have fractured a fragile venture, and could have given the public the impression that the design and process was flawed. The focus on XML-DEV therefore helped the credibility of the XML effort.

Initially I adopted the W3C model completely—including flaws—because I was impressed by the ability of warring companies to come together and collaborate. We desperately need that model in chemistry! And I ruled out discussion of the flaws. Remember also that the initial editorial group was a close-knit group of ex-SGML experts who had committed a substantial part of their waking lives to making XML work! Later, after XML had "succeeded" and as more W3C activities depended on it, discussion emerged that showed that there were tensions, and that poor design might creep into drafts for political reasons. So discussion became more general.

Moving Home

OASIS, a non-profit international consortium, was founded in 1993. Its aim is to spread the adoption of product-independent formats based on public standards like SGML and XML. I asked why Professor Murray-Rust chose to move the list to OASIS after three years at Imperial College:

When I set up the list I persuaded Henry to manage it on my—and your—behalf. Henry is interested in XML as a means to an end (chemistry, and more generally electronic publishing) and his daily management of the tedium of the list is remarkable. Every day brings tens of error messages to his mailbox and weekends are worse. We had a fragile arrangement—running the list on a server in Imperial College to which we had no access. Responsibility without power! I got great fulfillment out of the list—at the expense of Henry's commitment

In mid 1999 Jon Bosak suggested we might transfer the list to OASIS and we accepted. The process is now finalized. The primary reason was to free Henry, but I also saw the need for trusted organizations in the information age. When I speak on e-publishing I emphasize the need to promote non-profit orgs and learned societies. So OASIS is a natural center for trust in this community, and the transfer is a small act of faith in that direction.

Things To Come

So, where next for XML-DEV? Are there new directions that the community might take? Professor Murray-Rust believes that the challenges are only just beginning:

The challenges that face us are enormous and will take years to solve. XML solves encoding and syntax precisely, and nothing else. In particular it does not solve semantics and ontology, but it highlights these as problem areas. I am worried that we shall get hardcoded solutions to these, partly because that is the "easiest" way and partly because it may be possible to capture markets. At best I see a "semantic and ontological marketplace" emerging—at worst, warfare.

XML-DEV is now the natural home for this. Much of the recent discussion has been on this topic and has been very valuable. I think it's critical that a communal approach to semantics is created because it isn't—or should not be—an area for competition. Otherwise we are back to "this message can only be understood with Plugh's library." I want to be able to send chemistry to anyone. And hardcoding the semantics will probably hardcode the ontologies. Given that the world has now effectively decided that XML is the transfer medium for information, XML-DEV must be proactive in developing semantics.

A specific instance of this is editing technology. If we (XML-DEV and others) can create a universal core for editing tools, then the e-publishing will truly shift to the masses . For example, Germany has a national initiative to produce electronic theses (in XML) and we need editing technology. There is enormous scope for communal Open Source activity here—others will know better how to develop it.

Henry Rzepa and Peter Murray-Rust should be applauded for their hard work. XML-DEV has played a key role in shaping today's XML community. It is lively and varied, receiving contributions from a wide spectrum of industries and applications. This assemblage of the technical community, from the "Desperate Perl Hacker" to the seasoned SGML veteran, is an important resource. Professor Murray-Rust summed it up well:

The key feature of XML-DEV is the meritocracy, mutual respect, and tolerance that has developed. OASIS must protect this. I and other list members will be vocal if not!

It looks like the next three years of XML community will still be in safe hands!