Tim Bray
Everyone from Forbes magazine to Microsoft Corp. CEO Steve Ballmer is
saying that XML is at the strategic center of our industry. That makes
me happy, as I helped invent XML. But there is a significant hype
problem, so let's stick with the basics.
XML is a set of rules for taking your data and encoding it in chunks of
text, which you can easily send across the Net and unpack at the
receiving end. It has good internationalization and error handling, and
nobody owns it.
That's all there is to it. It may sound simple, but you need something
like XML in any modern enterprise application -- whether you're
integrating your catalog lookup system with your extranet or building a
next-generation network GUI like my company's (Antarcti.ca) Visual Net
product.
XML isn't just a data transmission tool; enterprises around the world
are using it to author and store data as well. They do this because
information is valuable and sometimes long-lived. In a world where
software goes through a generation every 18 months and you can't
predict what you'll want to do tomorrow with today's data, XML is a
safe haven.
Consider the world of networking: In 1985, if you wanted to get
networking software, you could use: DECNet (DEC only), SNA (IBM only)
and so on. Today, if anyone introduced a new single-vendor networking
product, they'd be laughed out of the market. One of the promises of
XML is that "proprietary data format" will soon sound as silly
as "proprietary networking product."
There are other reasons to author and store XML. Before too long,
you're going to need to deliver data not just to Web browsers but also
to PDAs and cell phones. This evolution will require some careful
design and advanced technology. But the expansion will be much easier
if the data starts out in a format, such as XML, that isn't predisposed
to the browser (or a particular brand of PDA or model of cell phone or
even print.)
As for the future: Despite the billions of dollars and lifetimes of
effort that have been poured into the Web, it looks and works about the
same as in 1995. This can't last, and XML has a big role to play in
several generational changes that are coming.
To buy performance we're going to have to start getting more mileage
out of our desktop machines--using them as more than dumb HTML
terminals. Next, we're going to want visual interfaces to network data,
just as we demand a desktop GUI for PC data. Finally, we desperately
need smarter search tools than the brute-force word-munchers vendors
give us. XML will be the smart data that puts the desktop back to work,
the interchange format that drives the network GUI and the meta-data
protocol that makes searching less painful.
But from the CIO's point of view, the lesson XML teaches is to treat
your data right, because it's worth more than you know. XML, for the
first time, makes this possible.
