Re: Followup: opinions on Search services

[Joe Shaw <joeshaw novell com>]

Hi,

On Tue, 2005-04-26 at 18:20 -0500, Manuel Amador wrote:
> We have only one server for all users.  We thus dramatically reduce the
> load on multiuser systems.

Like I said before, I think you're going to end up spending a ton of
your time making sure you have the access controls correct, you'll need
to get pretty comprehensive security reviews before people ship it, etc.

For Beagle our target has been at the more traditional single-user
desktop.  We do some throttling based on the load of the machine so that
in the case of many users we won't bring the system down, but that does
mean that indexing of data on everyone's system will be slower.  On the
other hand, generally this is only an issue the first time you ever run
beagle and briefly at login time, so this is a tradeoff we're willing to
make.

> That is very good, I grant you that.  But there's no way we could
> intersect Lucene results with our own metadata search results, such as:
> 
> - documents created by rudd-o the last month that contain the word
> "paloma" and were written in plain text
> 
> .. unless we could move completely to Lucene, which you appear to have
> done.  By the way, does Lucene have pybindings?

Everything we search on we store in Lucene.  We also use extended
attributes with an sqlite database fallback for some meta information
like the last time we indexed the file.  This is also how we plan to do
data relationships in the future.

Lucene is written in Java, so I suspect you could use Jython to access
it.  We're using a port to C#, so you could use IronPython to use that
one.  But there isn't a direct port of Lucene to Python, no.

Joe