[
https://issues.apache.org/jira/browse/LUCENE-1187?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12571939#action_12571939
]
Eks Dev commented on LUCENE-1187:
---------------------------------
Paul, I think there is one CHEKME in DisjunctionSumScorer I have stumbled upon
recently when I realized
(token1+ token2+) query works way faster than (token1
token2).setMinimumSholdMatch(2). It is not directly related to the LUCENE-584,
but just as a reminder.
also I think there is a hard_to_detect_small_maybe_performance_bug in
ConjuctionScorer, :
{code:java}
// If first-time skip distance is any predictor of
// scorer sparseness, then we should always try to skip first on
// those scorers.
// Keep last scorer in it's last place (it will be the first
// to be skipped on), but reverse all of the others so that
// they will be skipped on in order of original high skip.
int end=(scorers.length-1)-1;
for (int i=0; i<(end>>1); i++) {
Scorer tmp = scorers[i];
scorers[i] = scorers[end-i];
scorers[end-i] = tmp;
}
{code}
It has not been detected so far as it has only performance implications (I
think?), and it sometimes works and sometimes not, depending on number of
scorers:
to see what I am talking about, try this "simulator":
{code:java}
public static void main(String[] args) {
int[] scorers = new int[7]; //3 and 7 do not work
for (int i=0; i<scorers.length; i++) {
scorers[i]=i;
}
System.out.println(Arrays.toString(scorers));
int end=(scorers.length-1)-1;
for (int i=0; i<(end>>1); i++) {
int tmp = scorers[i];
scorers[i] = scorers[end-i];
scorers[end-i] = tmp;
}
System.out.println(Arrays.toString(scorers));
}
{code}
for 7 you get:
[0, 1, 2, 3, 4, 5, 6]
[5, 4, 2, 3, 1, 0, 6]
instead of [5, 4, 3, 2, 1, 0, 6]
and for 3
[0, 1, 2]
[0, 1, 2] (should be [1, 0, 2])
> Things to be done now that Filter is independent from BitSet
> ------------------------------------------------------------
>
> Key: LUCENE-1187
> URL: https://issues.apache.org/jira/browse/LUCENE-1187
> Project: Lucene - Java
> Issue Type: Improvement
> Reporter: Paul Elschot
> Priority: Minor
>
> (Aside: where is the documentation on how to mark up text in jira comments?)
> The following things are left over after LUCENE-584 :
> For Lucene 3.0 Filter.bits() will have to be removed.
> There is a CHECKME in IndexSearcher about using ConjunctionScorer to have the
> boolean behaviour of a Filter.
> I have not looked into Filter caching yet, but I suppose there will be some
> room for improvement there.
> Iirc the current core has moved to use OpenBitSetFilter and that is probably
> what is being cached.
> In some cases it might be better to cache a SortedVIntList instead.
> Boolean logic on DocIdSetIterator is already available for Scorers (that
> inherit from DocIdSetIterator) in the search package. This is currently
> implemented by ConjunctionScorer, DisjunctionSumScorer,
> ReqOptSumScorer and ReqExclScorer.
> Boolean logic on BitSets is available in contrib/misc and contrib/queries
> DisjunctionSumScorer calls score() on its subscorers before the score value
> actually needed.
> This could be a reason to introduce a DisjunctionDocIdSetIterator, perhaps as
> a superclass of DisjunctionSumScorer.
> To fully implement non scoring queries a TermDocIdSetIterator will be needed,
> perhaps as a superclass of TermScorer.
> The javadocs in org.apache.lucene.search using matching vs non-zero score:
> I'll investigate this soon, and provide a patch when necessary.
> An early version of the patches of LUCENE-584 contained a class Matcher,
> that differs from the current DocIdSet in that Matcher has an explain()
> method.
> It remains to be seen whether such a Matcher could be useful between
> DocIdSet and Scorer.
> The semantics of scorer.skipTo(scorer.doc()) was discussed briefly.
> This was also discussed at another issue recently, so perhaps it is wortwhile
> to open a separate issue for this.
> Skipping on a SortedVIntList is done using linear search, this could be
> improved by adding multilevel skiplist info much like in the Lucene index for
> documents containing a term.
> One comment by me of 3 Dec 2008:
> A few complete (test) classes are deprecated, it might be good to add the
> target release for removal there.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
