Tim Bray, one of XML's creators, talks about HTML's successor

Interview by Wolf-Rüdiger Hansen - Director of IT Press Relations at Software AG;
extract from Software Report no. 47, March 1999

The Web is the greatest system ever invented for delivering information to humans. Spanning the globe, it has become a standard means for consumers and enterprises to access, transmit and distribute all kinds of information. This information is based on the use of HTML (hypertext markup language), which has been a huge success. But the limits of HTML have become visible and a new standard is on the rise: XML (extensible markup language).

Tim Bray is one of the fathers of XML and co-editor of W3C's (World Wide Web Consortium) XML specification. W3C is commited to designing open Web standards. In this interview, Tim Bray discusses the future of XML.

SR: Why do we need XML?

The Web today is all about communicating to humans, which is important. But for the next phase, including electronic commerce, we need machine-to-machine communication, and we need it to be reliable, efficient, and internationalized. That's what XML is all about.

SR: Is XML a programming language?

No. XML is a data format. One of the main points of XML is that it is equally usable with any programming language, although modern object-oriented languages like Java seem to be an especially good fit. XML is also equally usable in any human language, because it's based on the Unicode standard for international texts, and equally usable on any operating system, because nobody owns it.

SR: What are the hard problems with XML?

XML is pretty easy to create, and pretty easy to display, but we have a big problem with storing and managing large quantities of it in a clean and efficient way. Therefore, I am very interested in Software AG's plans to deliver a DBMS based on a pure XML engine. I think that an increasing amount of business information will be based on XML so we'll be needing this kind of repository technology.

SR: Which role will all the existing database products play with respect to XML?

Most commercial databases of today are based on the relational SQL model. This is a wonderful model for the rows and columns of business data, but XML has a much more complex structure that includes sequence, hierarchy, and hypermedia. We have already learned that it's very difficult to deal with XML in a relational repository. I think that there is a huge opportunity for a new generation of database management systems that are designed for XML.

SR: Will XML databases and conventional SQL databases coexists for a long time or will they be substituted by XML products?

Obviously, the arrival of XML won't make existing databases go away. Once we've solved the hard problem of how to store and manage XML well, we have to solve the other hard problem of how to integrate XML with our existing databases. I am not sure it's going to be possible to give a useful relational view of XML data, but it certainly would be nice to have an XML view of relational data.

SR: What are the existing SQL DBMS suppliers doing about XML?

The only real announcement is from Oracle, in their 8i product. They use an architecture where part of the XML goes into tables and part of it goes into Context, their full-text engine. We'll have to wait and see how this works out in practice, but it makes me a bit nervous that the data is split into two repositories and this is user-visible. I'm expecting that IBM will do something in this area; they have been putting huge amounts of work into Java, and they seem to be very committed to XML.

SR: Do we need to worry about Microsoft dominating the XML world the way it dominates the desktop?

Yes, of course. Microsoft is smart and fast and very, very rich. If they can see a way to take control of the XML world I am sure that they will do it. Fortunately, I think that will be quite difficult, since XML is quite simple and is 100 percent open. Anybody with brains and energy has a chance to play. I'm also happy to report that the World Wide Web Consortium, the organization that sponsored the development of XML, has a process that seems quite successful at balancing the interests of all the users and vendors.

SR: When we have XML databases, how will we query them? Can SQL do the job?

I think we need a new XML-oriented query language. SQL as it stands today is more or less completely useless with XML. I suppose there is some hope with the extension facility in SQL3 of building an XML query module. But I am not optimistic. SQL is already insanely over-complicated, and one of the main advantages of XML is that it's intrinsically simple; this is why the market opportunity is so big.

Tim Bray is an independent consultant based in Vancouver, British Columbia, Canada. For more information, see www.textuality.com. Tim Bray can be contacted at tbray@textuality.com.