Be sure to realize that even with reverse wildcard support, the user can add
a trailing wildcard as well (double-ended wildcard) and then you are back in
the same boat.
The overall idea is that: 1) Hardware is much faster than just 3 or 4 years
ago, and 2) even though document counts are getting much larger, the number
of unique terms (which is all that matters for wildcard performance) does
not tend to grow as fast as document count grows. And, some fields have a
much more limited vocabulary (unique terms), so a leading wildcard is not
necessarily a big performance hit.
Technology advances. We should permit our mindsets to advance as well.
-- Jack Krupansky
-----Original Message-----
From: François Schiettecatte
Sent: Monday, November 12, 2012 2:38 PM
To: solr-user@lucene.apache.org
Subject: Re: Is leading wildcard search turned on by default in Solr 3.6.1?
John
You can still use leading wildcards even if you dont have the
ReversedWildcardFilterFactory in your analysis but it means you will be
scanning the entire dictionary when the search is run which can be a
performance issue. If you do use ReversedWildcardFilterFactory you wont have
that performance issue but you will increase the overall size of your index.
Its a tradeoff.
When I looked into it for a site I built I decided that the tradeoff was not
worth it (after benchmarking) given how few leading wildcards searches it
was getting.
Best regards
François
On Nov 12, 2012, at 5:33 PM, johnmu...@aol.com wrote:
Hi,
I'm migrating from Solr 1.2 to 3.6.1. I used the same analyzer as I was,
and re-indexed my data. I did not add
solr.ReversedWildcardFilterFactory to my index analyzer, but yet leading
wild cards are working!! Does this mean it's turned on by default? If
so, how do I turn it off, and what are the implication of leaving ON?
Won't my searches be slower and consume more memory?
Thanks,
--MJ
