I understood the issue to be complexity, not a lack of compelling
We'd LOVE to have the index available as an API -- it'd give us much
better control of process, and most importantly, would allow us to
figure out where the natural breaking points are for very large indexes.
Right now, we do some sort of black magic formula to calculate when an
index gets too big, and it's a real pain.
Having it as part of an API would also allow us to give more feedback as
to how long something is going to take -- our indexing and merging is
often on the order of 24 hours, so development is quite a challenge (and
can be very frustrating when we reach certain limits).
Anyway, I do hope the reason it's not available is because of the
complexity, not the lack of compelling arguments!
[mailto:email@example.com] On Behalf Of Peter Karman
Sent: Tuesday, May 24, 2005 11:45 AM
To: Multiple recipients of list
Subject: [SWISH-E] Re: C-API interface question
yes, the API (both C and Perl) is for libswish-e, which is just for
There is no plan to provide an API for indexing. We've talked about this
on the list before, and amongst the developers, and haven't yet heard a
compelling argument to have an indexing API. As Bill said it once, the
swish-e binary \is\ the API.
John Paige wrote on 05/24/2005 10:29 AM:
> I see that C-API interface of Swish-e only supports seraching. It
> does not support the C-API to index any file.Is it true? If so, is
> there any near term plan to have one?
Peter Karman . http://peknet.com/ . peter(at)not-real.peknet.com
Received on Tue May 24 08:50:23 2005