Big Bug - Lucene Query

Matt

unread,

Jun 25, 2010, 12:11:15 AM6/25/10

to ravendb

Query a lucene index (from client API or from web UI)

Dont worry about field names, just run on Raven/DocumentsByEntityName

3 fields
X:[[a]] AND Y:[[b]] AND Z:[[c]] - okay, returns 0 - Expected

Now try 4 fields:
X:[[a]] AND Y:[[b]] AND Z:[[c]] AND P:[[a]] - error parsing

What is going on????

Matt

unread,

Jun 25, 2010, 12:14:03 AM6/25/10

to ravendb

It seems the default limit for boolean clauses is 1024 so I doubt its
that.

This is gonna kill my day...

Matt

unread,

Jun 25, 2010, 12:53:07 AM6/25/10

to ravendb

Arrghh.. still no luck. I tried moving back to Analyzed but on
Analazyed when I do

mclass:"IS"

it also matches mclass:"GX", mclass:"RX" too..

So need to stick with exact string matching (including small strings
and whitespaces) so unless Im missing something NotAnalyzed along with
[[val]] is the way forward, but the boolean thing damn damn damn...

Matt

unread,

Jun 25, 2010, 1:03:36 AM6/25/10

to ravendb

hmmm.. I think this might have something to do with IS value as the
inverse searching for GX etc.. is okay

Stephen McKamey

unread,

Jun 28, 2010, 1:24:41 PM6/28/10

to ravendb

Matt,

I'm curious if the changes in this fork,

http://github.com/mckamey/ravendb

fix your issues? It sounds like you are running up against the way the
server-side parses NotAnalyzed values. I've made some changes to this
which does less pre-processing and hands off more of the parsing to
Lucene which should allow for far more complex queries.

I'd be interested to know one way or another if this fixes it for you.

Thanks,
Stephen

Ayende Rahien

unread,

Jul 1, 2010, 2:34:10 AM7/1/10

to rav...@googlegroups.com

This should be fixed now

Reply all

Reply to author

Forward