Python Secret Weblog for 2005 April 9 (entry 0)

(5) Sat Apr 09 2005 21:51 lxml 0.5.1 released:

I've just released version 0.5.1 of lxml, the Pythonic binding to the libxml2 and libxslt libraries. This because I got feedback to 0.5 which pointed to a critical bug in the way unicode was handled. This kind of feedback is why I should've released lxml long ago!

lxml 0.5.1, and more information, can be found here:

http://codespeak.net/lxml

Comments:

Posted by Paul Boddie at Tue Apr 12 2005 09:56

It looks very nice indeed! All it needs now is support for stylesheet parameters and extension functions, and I'll probably drop my own experiments with libxml2 altogether. Great work!

Posted by Martijn Faassen at Tue Apr 12 2005 12:45

Feel free to contribute! Stylesheet parameters shouldn't be too complicated, and I gained lots of experience with stylesheet extension functions when I looked into XPath extension systems. The APIs for the XSLT extension functions are apparently different (I haven't studied them yet), but we should be able to use the same principles in the API, I hope.

There is of course, more to do (what am I doing, writing shortcomings in an article promoting lxml??). One area in particular that needs more attention is a proper handling of libxml2 warnings and errors. When validation fails or a stylesheet fails to load, now you only know this, but not what went wrong. libxml2 tells you far more (just in a very C-ish manner, not very pretty). This should be turned into Pythonic (exception) information somehow.

Posted by Paul Boddie at Tue Apr 12 2005 14:36

This might be an excuse to get into Pyrex, I suppose. The main pitfalls (and peeves) right now are as follows: stylesheet parameters in libxslt are often treated like XPath expressions, so quoting is sometimes necessary; the existing Python bindings for XSLT extension functions use libxml2mod - you don't even get the usual node objects; libxml2 has an unsuppressable warning about "DAV:" URLs - grrr!

I was hoping that you just hadn't gotten around to putting this stuff in, but I really can't complain. ;-)

Posted by Paul Moore at Tue Apr 12 2005 15:35

Any chance of Windows binaries?

Posted by Martijn Faassen at Tue Apr 12 2005 16:53

There's a chance of Windows binaries if someone contributes them. I'd be very happy if this happened! I'd like to have them myself, but do not have the toolchain, experience nor time to do it myself, unfortunately.