Re: packages.debian.org
On Wed, 20 Jan, 1999, jmlb2@hermes.cam.ac.uk wrote:
> Given that from your description swish++ sounds like a general purpose
> indexer, which has been set up to index 'natural language' is it the best one
> for our purposes?
>
Once I removed a few conditions for removing words from indexing that weren't
appropriate for us, the system works quite well.
The index file for the Package web pages alone is about 6.5M and indexes
over 8000 files, I don't think that a simplistic search system will work
very well on something this large and this popular (swish++ is very fast).
Of course, this pales in comparison to what I have planned next. If I can
work out a few details, I may use swish++ for searching on the mailing
list archives (~1GB and > 160k files).
Jay Treacy
Reply to: