Published on O'Reilly (http://oreilly.com/)
 See this if you're having trouble printing code examples

Python Conference: Developers' Day

The Python Conference is now over. Much praise to the staff at Foretec Seminars who ran a satisfying and smooth conference in the midst of a blizzard that shut down the mid-Atlantic Seaboard. The conference ran like clockwork, and everybody--staff, presenters, and attendees--was pleasant, polite, and cooperative. How often does that happen in Washington D.C.?

New Python Releases

The big news on Developers' Day was that Guido announced the next releases of Python, called 1.6, to come out soon, probably in Summer 2000. (The current release is 1.5.2.) The purpose of this new point release is to get existing code contributions into the new code base. The major addition will be Unicode support, something the XML community has been seeking for some time. This is a major step forward for Python. In addition, Guido will include some existing bug fixes and probably Fredrik Lundh's new regular expressions engine. There may be other additions as well, but nothing that is not ready now will make it into this release. An alpha version of this release will probably be ready in March. Feature freeze will be "very soon".

Guido added this release because Python language development had slowed since he had begun working on Computer Programming for Everybody (CP4E). The original goals of 1.6 were too ambitious and have stretched out the release date. Consequently, the pre-Version 2 enhancements that don't make it into 1.6 will go into an additional point release, 1.7, for which a date is not currently available.

The next major version of Python is now being informally called Python 3000 or P3K. ("Your users will be surprised and pleased," someone said, "when you deliver it 300 years early.") It currently has a goal of a 2001 release. The contents of this version are still under discussion, but will almost certainly be backward-incompatible. A lot of computing work has happened since the release of Python 1.0.

Internationalization SIG

I attended the meeting of the Internationalization SIG. The discussion was a bit too technically detailed for me to report here. There was much interest in the addition of Unicode to Python in 1.6. Several attendees, however, felt that Unicode was insufficient to support legacy (pre-Unicode) applications. In Japanese, there are many more characters than those represented in Unicode. Furthermore, some Unicode characters are compromises among the characters of several different languages and therefore unsatisfying to many users.

This meeting benefited from the attendance of Dick Hardt from ActiveState. ActiveState announced at the conference that it will provide products and services for Python as well as Perl. In fact, Dick announced that he had hired two of O'Reilly's Python authors: David Ascher (co-author of Learning Python) and Mark Hammond (co-author of Python Programming on Win32). Dick and ActiveState have faced a lot of these Unicode and Asian character-set issues with Perl and offered some advice based on his experience.

It was also gratifying to see that the members of this SIG recommend Ken Lunde's book, CJKV Information Processing, as the key book for this technology area.

Documentation SIG

I was unable to attend this meeting, but I heard from Greg Wilson and Fred Drake that a fair amount of the discussion revolved around using the XML DTD DocBook as the basis of future documentation. (In the interest of fairness and shameless self-promotion, I should reveal here that I was the editor of Norm Walsh's and Len Muellner's book DocBook: the Definitive Guide.) The discussion broke down into two camps, as such discussions do:

Both views, in my opinion, are correct. Some group should define a reduced DocBook tag set appropriate to erstwhile technical authors. The complexity, if necessary, could be added after the documentation is done and incorporated. Get the information first; worry about formatting later. But the advantages of using a DTD that was designed primarily to handle just this kind of information are too valuable to pass up.


I followed the discussion in this SIG only with great difficulty. I can report only that there was much interest in the addition of Unicode in the 1.6 code base. The remaining issues were too detailed for me to do justice to them.

JPython SIG

JPython is a variation of the Python language that supports the Java environment. (In the JPython SIG, they refer to Python as CPython.)

Barry Warsaw announced that JPython 1.1 was released on the first day of the conference. How's that for timing?

The major problems for JPython currently have to do with various JVMs (in what Barry called Java's "Write Once Debug Everywhere environment") and some incompatibilities between the JPyton compiler and the JIT.

JPython 1.2 hopes to deliver stability, performance, and some new features like collections and support of Java security. (Security was felt by many of the SIG members to be very important.) Further out, JPython will support NetBeans integration and Java scripting frameworks.

One last issue was how JPython could capture the hearts and minds of Java developers. I floated the idea of a JPython Cookbook, written by the JPython community, with the Java developer community as the reviewers and audience. Not surprisingly, Barry Warsaw has been working with O'Reilly on such a book, and a barebones outline exists already. There was some interest in this idea and some discussion about which JPython features such a book might include.

Goodbye: See You in November

The goal of the Python Software Activity is to have another conference in November of this year, rather than next January, probably in Washington again. Certainly, after this conference's weather, there is much support for a kinder, gentler Washington climate. I expect to be there, and if attendance increases as much as it did this year, many of you will be there, too. See you.

Read about Day One and Day Two.

Return to: Frankly Speaking

Copyright © 2009 O'Reilly Media, Inc.