Xapian discuss - Oct 2005 - Python interface problems

Hello,

I'm trying to use xapwrap to index some documents in Xapian, but I get 
the following error.  Any ideas how to fix/circumvent this?  I find 
Xapian the best solution for my problem (indexing 20 million forum 
posts, messages, etc.).

Also, how 'fast' is Xapian?  I want to use it on a heavily loaded site, 
and I would probably generate 10-20 queries per second (but the queries 
would be limited to 1-8 words from a subset of 100-1000 words).  Would 
that work fine on a dual xeon with SCSI RAID 1 drives?

Here's the traceback:

Traceback (most recent call last):
  File "xapsearch.py", line 72, in ?
    _do_index('/home/test/dupa')
  File "xapsearch.py", line 61, in _do_index
    id.index({'ids': ["bulletin", i] + _gentags(),
'data': 'bulletin%d'
% i})
  File "xapsearch.py", line 39, in index
    self.indexer.index(doc)
  File "/home/test/xapwrap/index.py", line 820, in index
    xapDoc = doc.toXapianDocument(self.indexValueMap)
  File "/home/test/xapwrap/document.py", line 177, in toXapianDocument
    d.add_term(term.value)
  File "/usr/local/lib/python2.4/site-packages/xapian.py", line 230,
in
add_term
    def add_term(*args): return _xapian.Document_add_term(*args)
NotImplementedError: No matching function for overloaded
'Document_add_term'

On Mon, Oct 03, 2005 at 01:53:09PM +0200, Wojtek Sobczuk
wrote:
> I'm trying to use xapwrap to index some documents in Xapian, but I get 
> the following error.  Any ideas how to fix/circumvent this?
Xapwrap is a third party wrapper around Xapian's python bindings, so
I can't help much there.  The author does read this list.
> Also, how 'fast' is Xapian?  I want to use it on a heavily loaded
site,
> and I would probably generate 10-20 queries per second (but the queries 
> would be limited to 1-8 words from a subset of 100-1000 words).  Would 
> that work fine on a dual xeon with SCSI RAID 1 drives?
That should be fine, assuming you've got plenty of RAM in the box.  Note
that dual CPU isn't particularly useful here - it's I/O which is
critical so fast disks and plenty of RAM to cache the Btree blocks.

You might want to consider using the flint backend in 0.9.2 - it's
a work in progress, but the version in 0.9.2 is proving very solid.
It's already in production use at Gmane and a couple of other sites!
Arjen from tweakers.net posted some benchmarks to the list a while
ago which were promising.

Cheers,
    Olly