Mailing-List: contact user-help@couchdb.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@couchdb.apache.org
Content-Type: text/plain; charset=us-ascii
Mime-Version: 1.0 (Apple Message framework v1278)
Subject: Re: CouchDB Lucene boost problem
From: Robert Newson <rnewson@apache.org>
In-Reply-To: <D5104AE467DD4E89AC4E638A3177FD4F@chillibean.tv>
Date: Wed, 20 Jun 2012 15:07:23 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <8F779074-E347-4528-B67B-FE760AF7D2BB@apache.org>
References: <0A30A158CD7D43E7BAAFD2B93A6B7BB7@chillibean.tv>
 <09A5FADF-C288-4D70-9369-995106E318ED@apache.org>
 <D5104AE467DD4E89AC4E638A3177FD4F@chillibean.tv>
To: user@couchdb.apache.org


There's index-time and query-time boosting, perhaps that explains it?


On 20 Jun 2012, at 14:35, Rory Franklin wrote:

> The error is now gone, which is great. Thanks so much for fixing that =
so quickly!
>=20
> Looking at the results of the search (with debug on for readability), =
it doesn't look like the boost is actually getting applied? I have a =
boost on another field (same name, but without the sort_ prefix) and in =
the BooleanQuery section of the search result it looks like this:
>=20
> entry_human_id:1220091,boost=3D1.0
> The value in the index is actually 1.5. Is it the case that the boost =
isn't taking effect or that my query doesn't manually boost that field?
>=20
> Rory
>=20
>=20
> On Wednesday, 20 June 2012 at 13:21, Robert Newson wrote:
>=20
>> anyway, I fixed it.=20
>>=20
>> You'll need to delete the index already built, though, and I'd love =
to hear back from you when you try it.
>>=20
>> B.
>>=20
>> On 20 Jun 2012, at 12:05, Rory Franklin wrote:
>>=20
>>> Hi,
>>>=20
>>> We've got a machine with couchdb-lucene 0.9 on it and various =
machines that use 0.7 and there seems to be a discrepancy between the =
two versions.
>>>=20
>>> In one of our indexes we are boosting a particular field so that it =
comes above others in search results (we are searching for an ID and =
linked ID's in other documents, but want the document that matches the =
ID directly to return above the linked documents), and the definition =
looks like this:
>>>=20
>>> ret.add(doc.entry_human_id, {'field':'sort_entry_human_id', 'type' : =
'int', 'boost' : 1.5})
>>>=20
>>> On 0.7 this works absolutely fine, but on 0.9 we are seeing this =
error:
>>>=20
>>> 2012-06-20 10:48:04,251 WARN [lia_development] Exiting due to =
exception.
>>> java.lang.UnsupportedOperationException: You cannot set an =
index-time boost: norms are omitted for field 'sort_entry_human_id'
>>> at =
org.apache.lucene.index.DocInverterPerField.processFields(DocInverterPerFi=
eld.java:85)
>>> at =
org.apache.lucene.index.DocFieldProcessorPerThread.processDocument(DocFiel=
dProcessorPerThread.java:276)
>>> at =
org.apache.lucene.index.DocumentsWriter.updateDocuments(DocumentsWriter.ja=
va:852)
>>> at =
org.apache.lucene.index.IndexWriter.updateDocuments(IndexWriter.java:2167)=

>>> at =
com.github.rnewson.couchdb.lucene.DatabaseIndexer.handleResponse(DatabaseI=
ndexer.java:393)
>>> at =
com.github.rnewson.couchdb.lucene.DatabaseIndexer.handleResponse(DatabaseI=
ndexer.java:83)
>>> at =
org.apache.http.impl.client.AbstractHttpClient.execute(AbstractHttpClient.=
java:735)
>>> at =
org.apache.http.impl.client.AbstractHttpClient.execute(AbstractHttpClient.=
java:709)
>>> at =
org.apache.http.impl.client.AbstractHttpClient.execute(AbstractHttpClient.=
java:700)
>>> at =
com.github.rnewson.couchdb.lucene.DatabaseIndexer.run(DatabaseIndexer.java=
:473)
>>> at java.lang.Thread.run(Thread.java:680)
>>>=20
>>>=20
>>>=20
>>> Removing the boost and restarting couchdb-lucene resolves the issue, =
I'm just curious as to what the problem may be?
>>>=20
>>>=20
>>>=20
>>> Thanks,
>>> Rory
>>>=20
>>=20
>>=20
>>=20
>=20
>=20