On Fri, Aug 01, 2003 at 12:48:06PM -0700, jb wrote:
> hi there,
> 
> i've seen that people ask for python wrappers every now and then -
> there's a python module wrapping the C API (release  2.4.0-pr1) here:
>    http://jibe.freeshell.org/bits/SwishE/
> 
> it is not too tested, but it looks like working, also there's a cgi script
> to search mailman archives in the distrib, example here:
>    http://sarai.indymedia.org/~jb/search.cgi).
> 
> comments are welcome.

Very cool.  So you are indexing pipermail archives?  Are you doing 
any special parsing of the docs -- such as extracting out email, name or 
subject?

If you are indexing mailing lists you might consider trying to use the 
-u option to update an index instead of reindex.  You would need to 
compile swish-e with --enable-incremental configure option.  That would 
allow you to add messages to the index as they arrive.  You would need 
to use some type of locking to allow only one update to happen at a 
time.

I have not tested it more than once with a very few documents, so 
consider it experimental.  But it would be nice to get some feedback on 
it.  It's really "update" not "incremental" because it only adds files 
to the index and doesn't remove files or update existing ones.  At least 
I think that's how it works.

Oh, and do you have any comments on the API?  Was it easy to understand 
and follow.  Documentation patches are always welcome!

Thanks,


-- 
Bill Moseley
moseley@hank.org