lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paul Elschot <paul.elsc...@xs4all.nl>
Subject Re: RLE Compressing bit vectors, just toughts
Date Sun, 05 Aug 2007 09:40:37 GMT
On Sunday 05 August 2007 10:35, eks dev wrote:
> 
...
> 
> Also, this way we are not going to reduce the number of VInt decoding  (by 
far the most popular method call in Lucene :) 
> 
> So far, my best bet would be to do it directly on index, every time you see  
a lot of deltas  == 1, you try to encode them with some 0xFF(run length). On 
decoding, next() and skipTo() should not be all that complex (I hope). Must 
see how it can cooperate with skipping info.

One way to reduce the complexity of next() and skipTo() is by keeping a buffer
of interim results, much like TermScorer does now, but without the frequency
information.

Regards,
Paul Elschot

---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-dev-help@lucene.apache.org


Mime
View raw message