lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Robert Muir <rcm...@gmail.com>
Subject Re: wildcardquery rewrite()
Date Wed, 07 Oct 2009 13:47:31 GMT
Mark, I am set up to do these tests with a large term dict. I will see
if there is any improvement.

In my opinion in general, even if the improvement is very small, if
its trivial to rewrite to a faster/simpler query, we should.
Future improvements to lucene might make the simpler query might
become even faster, etc.

On Wed, Oct 7, 2009 at 8:08 AM, Mark Miller <markrmiller@gmail.com> wrote:
> bq. I don't think the prefix enumeration is really that much faster than
> the wildcard one,
>
> We should do some tests. If it is much faster, this would be a nice
> optimization. I think it could be worth it when matching a
> lot of terms - never tested though.
>
> Robert Muir wrote:
>> separately, perhaps we should consider doing the prefixquery rewrite
>> here for wildcardquery.
>>
>> for example, SolrQueryParser will emit these 'wildcardqueries that
>> should be prefixqueries' if you are using the new reverse stuff for
>> leading wildcards: WildcardQuery(*foobar) ->
>> WildcardQuery(U+0001raboof*)
>>
>> I don't think the prefix enumeration is really that much faster than
>> the wildcard one, but still thought I would mention it.
>>
>> On Tue, Oct 6, 2009 at 10:22 PM, Robert Muir <rcmuir@gmail.com> wrote:
>>
>>> someone asked this question on the user list:
>>> http://www.lucidimagination.com/search/document/6f38de391b242102/prefixquery_vs_wildcardquery
>>>
>>> it made me look at the wildcard rewrite(), where i see this:
>>>    if (!termContainsWildcard)
>>>      return new TermQuery(getTerm());
>>>
>>> is it a problem the boost is not preserved in this special case?
>>>
>>> is it also a problem that if the user sets the default MultiTermQuery
>>> rewriteMethod to say, CONSTANT_SCORE_FILTER_REWRITE,
>>> that this rewritten TermQuery isn't wrapped with a constant score?
>>>
>>> Sorry if it seems a bit nitpicky, really the issue is that I want to
>>> do the right thing for a more complex query I am working on, but don't
>>> want to overkill either.
>>> --
>>> Robert Muir
>>> rcmuir@gmail.com
>>>
>>>
>>
>>
>>
>>
>
>
> --
> - Mark
>
> http://www.lucidimagination.com
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: java-dev-unsubscribe@lucene.apache.org
> For additional commands, e-mail: java-dev-help@lucene.apache.org
>
>



-- 
Robert Muir
rcmuir@gmail.com

---------------------------------------------------------------------
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