Tim Bray | No easy road to interoperability

After he helped develop the Extensible Markup Language, Tim Bray could have rested on his laurels, collecting the occasional consulting check. Instead he continued to innovate.From 1996 to 1998, Bray co-wrote version 1 of the XML specification, which subsequently made its way into countless Web applications and other networking software. Today he is the director of Web technologies for Sun Microsystems Inc. He also co-chairs the Atom Publishing Format and Protocol Working Group, an Internet Engineering Task Force body that is laboring to standardize a data feed format for Web pages, based on XML.Despite his expertise in IT, Bray actually started out in another field altogether'rock and roll. While pursuing a double major in mathematics and computer science at the University of Guelph in Ontario, he organized hundreds of rock and roll concerts. He also put on plays and other theatrical performances in the Toronto exurb. After college, however, he was scooped up by Digital Equipment Corp., where he worked as a software specialist. He later co-founded content management software provider Open Text Corp. GCN associate writer Joab Jackson caught up with Bray by phone.BRAY: In 1996, it was obvious that the Web was going global. The Web was all based on the [Hypertext Markup Language]. HTML is a wonderful language for displaying information to people. However, HTML is not designed for machine-to-machine interchange of data.A lot of us were getting pretty excited by using the Web infrastructure to build business applications that could do machine-to-machine transactions. I was working on large-scale publishing systems, things like Boeing Co. technical documentation and the legislation of the European Parliament. The Standard Generalized Markup Language was designed specifically for this kind of machine-to-machine stuff, but it never became widely adopted because it was too complicated. The answer seemed pretty clear: You take SGML, throw away the 90 percent of complicated bits that people didn't use and add a few enhancements to make it Web friendly. And that's what we did. It wasn't a big job. There were 11 of us on the committee and it only took us 10 months to do the core design.BRAY: Semantic interoperability was never one of the primary goals of XML. XML is, by and large, pretty easy, whereas semantic interoperability is insanely difficult.Anyone who has spent years in the IT business quickly learns that semantic capability is a horrible, horrible set of problems. You would think that something like a 'part number' or a 'ship date' would be a simple concept, but you quickly discover that it is not. Two departments in the same company will each have a radically different idea of what a 'ship date' is.There is a recently finalized set of standards called Universal Business Language. It is a basic set of business documents like invoices and purchase orders. It has taken them 10 solid years of work [just] to get UBL nailed down in a generally satisfactory way. So let's not kid ourselves that there are any quick, easy solutions in the interoperability space.BRAY: RDF is a general schema for exchanging metadata. By metadata we mean assertions about things. The important thing about RDF is its data structure model. Everything is a three-part object, where you say this project has this property with that value. That enables you to build some pretty sophisticated data structures. The fact that RDF is written in XML is nice, but it is really not essential to RDF.Now some really smart people, including Tim Berners-Lee, inventor of the Web, think there is a huge value-add to be had with RDF. I'd be inclined to take those opinions seriously, but there is not an existing deployed success story just yet. RDF remains largely unproven.BRAY: At the moment, RSS is the single most successful application of XML. There are substantially [more than] 10 million RSS feeds in the world now, and the number is growing at like 10,000 a day. The growth figures look very much like the early days of the Web. My personal belief is that RSS will continue to grow and will be thoroughly mainstream within a few years.Having said that, there was a recent Nielsen study that showed that only 11 percent of Web users are actually using RSS. So there is a huge opportunity for growth. We still have a problem with subscription. We don't really have one-click subscription yet. You see these little red XML buttons on Web sites, but to use them, you have to copy the [Web address] and drop it into your aggregator. But by this time next year, all the popular browsers will have one-click subscriptions.RSS is already highly used in the intelligence community. I'm somewhat connected with the intelligence community [and have heard of] lots of wildly successful RSS deployments, obviously running on intranets where we can't see them. It is an extremely efficient way to deliver time-sensitive information that arrives at unpredictable intervals.BRAY: RSS works very well for what it does, but there are some problems with it'geeky technical problems that make systems administrators' lives harder. Probably the biggest problem with RSS is that there are nine different versions out there and there is not much chance of them being unified. None of them have been through a standards organization.So Atom is the attempt to get together, write down the parts we know that work, get them imprinted by the standards organizations, so there is an official stack that you can point to.BRAY: XML is a textual data format. One of the great virtues of XML is that it is profoundly independent of any operating system or programming language. And that is one of its selling points'it is extremely successful in stitching together enterprise applications across the network.When I say I will give you XML, that is a very clear statement that I will send you a stream of text with angle brackets in it, and that is an important part of the value proposition that we want to be very careful about weakening. There's no doubt at all [that] there are lots of people who interchange XML, but once they receive the XML, they immediately turn that into a binary format and ship it around their system.I don't think there is anything wrong with people turning XML binary. The question is, should there be an official binary encoding that is labeled as 'Binary XML'? The World Wide Web Consortium is looking pretty intensely at that. A lot of us are pretty dubious at that idea. You'd like it to be smaller, easier to parse and be self-describing in the same way that XML is. But it is far from clear that you can have one binary encoding that will meet all these objectives.BRAY: John Prine, Leon Redbone, Billy Preston, the New York Dolls, Blue Oyster Cult, Billy Joel, Rush, John Cale, Maria Muldaur. They all tend to blur together.Many people think working in the technology business is intense and has a lot of emergencies. Hah! After you've dealt with a few sobriety-challenged performers, a few dozen belligerent roadies and a few thousand drunken fans, there's nothing scary about high-tech.