developer at csrules.dyndns.org
Thu Jun 24 20:53:32 PDT 2004
Andrzej Bialecki wrote:
> I'm coming here from the Java Lucene camp - just spotted the
> announcement, and I'm quite excited. This is an excellent news, and
> the decision to split it as a separate project is very wise, IMHO. You
> can expect other language bindings to pop up very quickly... :-)
> I got the Windows binary, initially I had some problems opening an
> index - it looks like the compound file format is giving PyLucene a
> trouble (I think it was already supported in the 1.3-final), but when
> I converted it to the classic format it groks it just fine. However,
> compound format is your friend (it is the default in 1.4 branch).
> I'll try to setup some comprehensive benchmark, if time permits - I'd
> appreciate your help, e.g. the scripts you guys are using to benchmark
> PyLucene. I noticed that many base classes from Lucene (like primitive
> Query types) are not exposed, you can only produce them through the
> QueryParser (but this has some unpleasant side effects). Is it
> possible to add them? As well as other types of analyzers - the
> StandardAnalyzer is quite intrusive, and its escaping quirks are close
> to black art...
> Thanks in advance, and congratulations on the excellent project!
AFAIK there's no comprehensive benchmark suite being used.
As for adding classes. I think what has been done is add the base of
what is needed by chandler. Other stuff that i've needed i've tried to
add myself and provide a patch or diff to the developers.
Would you mind elaborating on what is wrong with the StandardAnalyzer I
dont' know enough about different quirks to know what would be
appropriate to use. It seemed to me that adding analyzers would be
pretty straight forward.
More information about the pylucene-dev