Thanks Hoss. I apologize for being slightly off topic. Instead of
using Solr, I am trying to steal from it, and so Lucene was still
mostly what was on my mind :) I posted too soon as well because I
finally found the wiki page on caching at Solr's site.
I have just implemented the "warm a new searcher" trick in my Lucene
app, and I am now looking into the best way to add caching--if caching
turns out to be beneficial to the app I am making. Of course I might
just drop all of this and use Solr, but it seems to be a nice learning
experience to go through all of this using just Lucene. Thanks for the
high level answers though...exactly what I was looking for.
- Mark
: It looks like Solr does not use a simple cached queryfilter as a query
: cache. Why is this? Is a cache queryfilter not efficient enough? Is this
: alternate method just so you can easily load the cache from an old
: Searcher to a new Searcher? Any info appreciated.
This would probably be better addressed on the solr-user mailing list --
please post any followup question there.
I'll through out some high level answers though in case any one else is
interested...
* there's nothign to stop a solr request handler from using
QueryFilter's if they want (i have and do)
* Solr has several types of caches for different things
* the "filterCache" has the most direct conceptual mapping to how a
QueryFilter and CachedWrapperFilter work. The big benefits it has are:
- the DocSet API allows for more memory efficient representations of
small sets of documents then just a BitSet
- it can be configured with a size and replacement strategy
(CachedWrapperFilter is more for things that can live "forever"
relative the IndexReader)
-Hoss
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
