[ANN]VTD-XML 2.10

[Stefan Behnel]

python at bdurham.com, 27.02.2011 13:52:
> How does VTD-XML compare to XML tools in the stdlib or to 3rd party
> alternatives like lxml?

For one, I'm not aware of any Python wrappers for vtd-xml, despite having 
seen lots of announcements by Jimmy on this list already (not the 
python-announce list, *this* list). So, IMHO, these announcements are to be 
considered spam, or at least vapourware. Also, Jimmy commonly does not 
respond to questions regarding his announcements, so I assume he doesn't 
actually read c.l.py but uses it solely to post his announcements.

That being said, vtd-xml is supposed to be fast, and there are benchmarks 
on the web that seem to suggest that this may be true. For example,

http://xmlbench.sourceforge.net/results/benchmark200910/index.html

lists its parser as being about twice as fast as libxml2, which is used by 
lxml.

I say "may be true", because benchmarks rarely indicate the performance 
behaviour in real world code. It does not seem completely unreasonable to 
me that vtd-xml is interesting for Java developers, where XML performance 
isn't really whopping cool and simple-to-use XML tools are rare anyway (I 
never tried it, but vtd-xml also has an aura of a not-so-easy-to-use tool).

It does not seem unreasonable that there are use cases for a very fast 
parser that justify the time it may take to get used to the tool. But its 
'different' nature also makes it clearly lack a lot of tooling around the 
actual parser, which prevents it from positioning itself as a general 
purpose XML tool. I'm yet to be convinced that vtd-xml is an interesting 
tool for a Python developer.

Stefan