Related link: http://conferences.oreillynet.com/os2002
One of the most gratifying tidbits I heard today at the
O’Reilly Open Source Convention
was not in the many tutorials but over breakfast, when I asked several
developers where the applications are that could catapult it
into popularity. (This was a follow-up to a conversation I had two days ago
at that time.)
Jabber is a lightweight protocol based on XML that potentially lets
people exchange structured data and bring automated programs to bear
upon it as easily as they now do instant messaging. Jabber is open, simple,
and wildly extendible–all winning qualities–and it could be the
basis of new, collaborative activities of great benefit to businesses
and online communities alike. But no one tags data just for the hell
of it; Jabber requires a killer app to show its power. I haven’t seen
announcements of anything promising.
The Jabber developers explained that interesting applications were
being developed by business in-house for their own use rather than
publicly. Enterprises are the target of many companies currently
pushing proprietary software and protocols; perhaps the simple
but flexible Jabber protocol will ultimately prove to be the solution
Simplicity was also the virtue claimed for Python by Keyton Weissinger
Python and XML
It seemed to me that the interface for various XML activities was
pretty similar to the interfaces offered by other languages, although
I did notice some nice convenience functions. Weissinger’s cocky
demonstrations of interactive programming, which often required a
couple rounds of corrections, showed that the interface
was not as simple as all that.
It was only later that I learned about Python’s value from Simon
St. Laurent, a fellow editor who edits books on XML and the
Web. (Simon operates his own
but gave me permission to quote him on this.) According to Simon,
Python’s object-oriented foundation fits neatly with XML, while its
loose typing makes programming less cumbersome than with a
strongly-typed language like Java.
I would think that one’s judgment of simple vs. complex depends less
on the programming interface that on one’s view of XML itself–if you
feel the specifications make sense, you probably feel comfortable with
programming it–and perhaps most of all with the fit between XML and your
particular data and application. Furthermore, if the text within a tag
requires a lot of fancy processing, your XML interface may be less
important than the text processing capabilities of your language.
I learned that structure can’t banish complexity in system
administration, either, when I stopped in on the session
OpenNMS: Managing your Networks, Systems, and Applications by
Shane O’Donnell. This impressive system provides a lot of choices for
monitoring, logging, and notification. But throw in an ill-considered
configuration option and you could be up all night. OpenNMS could act
like the guards in Monty Python and the Holy Grail who
refused to generalize or apply common sense to their master’s
directions. Used carefully, it looked like a big time-saver.
Nobody called Perl simple, but Larry Wall did promise in his annual
State of the Onion address that it would get simpler in Perl 6. His
half-hour talk was admirably designed and delivered, considering that
it must have been prepared in the interval since the appearance of the
August 2002 issue of Scientific American
on which Larry based it.
The evening ended with Jon Orwant’s notorious Internet Quiz Show, which the defending champions won for the third year running. Perhaps the Internet Quiz Show is popular because the loony randomness of the questions reflects the day-to-day unpredictability of the typical programmer’s or sysadmin’s life. Who can tell what nonconformant server or unrevealed code bug will interrupt one’s work with a crisis at any moment?