You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@lucene.apache.org by John Wang <jo...@gmail.com> on 2009/04/25 05:34:02 UTC
perf enhancement and lucene-1345
Hi Guys:
A while ago I posted some enhancements to disjunction and conjunction
docIdSetIterators that showed performance improvements to Lucene-1345. I
think it got mixed up with another discussion on that issue. Was wondering
what happened with it and what are the plans.
Thanks
-John
Re: perf enhancement and lucene-1345
Posted by John Wang <jo...@gmail.com>.
Michael:
Actually this is not what I meant.
It was discussion that was wrapped in Lucene-1345 where I posted some
optimizations for Conjunction and Disjunction Scorers and some numbers.
Because of the confusion of what Lucene-1345 is, the loop was never closed.
I am just writing to follow up.
Thanks
-John
On Mon, Apr 27, 2009 at 10:14 AM, Michael McCandless <
lucene@mikemccandless.com> wrote:
> I think you mean this thread:
>
> http://markmail.org/message/idgcnxmbyo3yjdiw
>
> right?
>
> I would love to see these in Lucene... P4Delta, which Paul has started
> under LUCENE-1410, is clearly a win, but is a biggish change to Lucene
> since all offsets would need to become "blockID + offsetWithinBlock".
> LUCENE-1458 (further steps flexible indexing) tries to make things
> generic enough that P4Delta can simply be a different codec.
>
> On the logic operators for combining DocIDSets... how do these differ
> from what we already do in BooleanScorer[2]? (I haven't had a chance
> to get a good look at Kamikaze yet).
>
> Mike
>
> On Fri, Apr 24, 2009 at 11:34 PM, John Wang <jo...@gmail.com> wrote:
> > Hi Guys:
> > A while ago I posted some enhancements to disjunction and
> conjunction
> > docIdSetIterators that showed performance improvements to Lucene-1345. I
> > think it got mixed up with another discussion on that issue. Was
> wondering
> > what happened with it and what are the plans.
> > Thanks
> > -John
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: java-dev-unsubscribe@lucene.apache.org
> For additional commands, e-mail: java-dev-help@lucene.apache.org
>
>
Re: perf enhancement and lucene-1345
Posted by Michael McCandless <lu...@mikemccandless.com>.
I think you mean this thread:
http://markmail.org/message/idgcnxmbyo3yjdiw
right?
I would love to see these in Lucene... P4Delta, which Paul has started
under LUCENE-1410, is clearly a win, but is a biggish change to Lucene
since all offsets would need to become "blockID + offsetWithinBlock".
LUCENE-1458 (further steps flexible indexing) tries to make things
generic enough that P4Delta can simply be a different codec.
On the logic operators for combining DocIDSets... how do these differ
from what we already do in BooleanScorer[2]? (I haven't had a chance
to get a good look at Kamikaze yet).
Mike
On Fri, Apr 24, 2009 at 11:34 PM, John Wang <jo...@gmail.com> wrote:
> Hi Guys:
> A while ago I posted some enhancements to disjunction and conjunction
> docIdSetIterators that showed performance improvements to Lucene-1345. I
> think it got mixed up with another discussion on that issue. Was wondering
> what happened with it and what are the plans.
> Thanks
> -John
---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-dev-help@lucene.apache.org