lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chong, Herb" <>
Subject RE: SQLDirectory
Date Fri, 06 Feb 2004 21:32:50 GMT
i agree mostly. however, i wonder if moderate sized indices that haven't been merged might
be able to realize adequate performance this way while at the same time allowing some of the
other desireable properties of using a database. what i am thinking of is that instead of
a single posting list for a term that there be several moderately large ones, something like
a set of indices from separate indexing runs before they are merged. delete performance would
still be lousy, but add performance might be OK if the add operations were done in memory
before committing them to the database. there would be a second index column, something like
index number or something like that.


-----Original Message-----
From: Doug Cutting []
Sent: Friday, February 06, 2004 4:28 PM
To: Lucene Users List
Subject: Re: SQLDirectory

That sort of representation works well for a static index, but it's not 
efficient to modify.  One could do this in Lucene by directly 
implementing the IndexReader API against an SQL database.  The database 
tables could be populated from a file-based IndexReader.  But I think 
implementing IndexWriter this way would be very slow.

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message