It is difficult to say for sure - unless somebody actually does a lot of
benchmarking tests with various distributions of data in the fields and
various field types (e.g., some are strings and some are text, and the
cardinality of the string values.) I would suspect that the two would be
roughly equivalent. I mean, if you search each field separately, that field
has only its subset of the data, and the copy field has essentially the sum
of the per-field subsets.
I would say that you should go with edismax "dismax" search (qf = list of
fields and boosts) unless you have a clear reason to go the other way.
-- Jack Krupansky
-----Original Message-----
From: Otis Gospodnetic
Sent: Wednesday, February 06, 2013 8:04 PM
To: solr-user@lucene.apache.org
Subject: Re: copyField vs single field
The latter, I believe, but you lose the ability to give different weights
to matches on different fields.
Otis
Solr & ElasticSearch Support
http://sematext.com/
On Feb 6, 2013 2:34 PM, "adm1n" <evgeni.evg...@gmail.com> wrote:
Hi,
Let's assume I have to search for a string (textField) in 6-7 different
fields (username, firstname, lastname, etc). Which one will have better
performance:
username:test OR firstname:test OR lastname:test
or defining some copyField and searching within it like somecopyfield:test
thanks.
--
View this message in context:
http://lucene.472066.n3.nabble.com/copyField-vs-single-field-tp4038832.html
Sent from the Solr - User mailing list archive at Nabble.com.
