You are viewing a plain text version of this content. The canonical link for it is here.
Posted to solr-user@lucene.apache.org by vybe3142 <vy...@gmail.com> on 2012/04/18 01:18:47 UTC

SOLR 4 / Date Query: Spurious Results: Is it me or ... ?

I wrote a custom handler that uses externally injected metadata (bypassing
Tika et all)

WRT Dates, I see them associated with the correct docs when retrieving all
docs:

BUT: 

looking at the schema analyzer, things look wierd:
1. Top terms = -1
2. The Dates are all mixed up with some spurious 1970 dates thrown in (I can
get rid of the 1970 dates if i use type "date" vs "tdate")
3. Multi Valued values (should only be one per doc, as per input data, even
though the schema allows it).

Any ideas what, if anything, I'm doing wrong?

See pic http://lucene.472066.n3.nabble.com/file/n3918636/Capture.jpg 

Here's my SOLR schema:




--
View this message in context: http://lucene.472066.n3.nabble.com/SOLR-4-Date-Query-Spurious-Results-Is-it-me-or-tp3918636p3918636.html
Sent from the Solr - User mailing list archive at Nabble.com.

Re: SOLR 4 / Date Query: Spurious Results: Is it me or ... ?

Posted by vybe3142 <vy...@gmail.com>.
Thanks for clarifying.

I figured out the (terms=-1). It was my fault. I attempted a truncate of the
index in my test case setup by issuing a delete query and think the
subsequent commit might not have taken effect by the time the subsequent 
index queries started.

--
View this message in context: http://lucene.472066.n3.nabble.com/SOLR-4-Date-Query-Spurious-Results-Is-it-me-or-tp3918636p3920652.html
Sent from the Solr - User mailing list archive at Nabble.com.

Re: SOLR 4 / Date Query: Spurious Results: Is it me or ... ?

Posted by Erick Erickson <er...@gmail.com>.
Your schema didn't come through, but...

1> why terms=-1 I don't know. I have a build from this
     morning and it's fine. When's yours?
2> date .vs. tdate. Yes, that's kind of confusing, but
     the Trie types inject some extra stuff in the field
     that allows the faster range queries, I think of it
     as "navigation data". These get displayed as
     1970 dates (e.g. the epoch). Ignore them.
3> I don't quite understand here. If you're still talking about
     a tdate field, could the "navigation data" account
     for it? That data shouldn't belong to any document and
     isn't really putting multi-values in any doc. Changing the
     schema type to not be multivalued should show this is the
     case if so.

Best
Erick

On Tue, Apr 17, 2012 at 7:18 PM, vybe3142 <vy...@gmail.com> wrote:
> I wrote a custom handler that uses externally injected metadata (bypassing
> Tika et all)
>
> WRT Dates, I see them associated with the correct docs when retrieving all
> docs:
>
> BUT:
>
> looking at the schema analyzer, things look wierd:
> 1. Top terms = -1
> 2. The Dates are all mixed up with some spurious 1970 dates thrown in (I can
> get rid of the 1970 dates if i use type "date" vs "tdate")
> 3. Multi Valued values (should only be one per doc, as per input data, even
> though the schema allows it).
>
> Any ideas what, if anything, I'm doing wrong?
>
> See pic http://lucene.472066.n3.nabble.com/file/n3918636/Capture.jpg
>
> Here's my SOLR schema:
>
>
>
>
> --
> View this message in context: http://lucene.472066.n3.nabble.com/SOLR-4-Date-Query-Spurious-Results-Is-it-me-or-tp3918636p3918636.html
> Sent from the Solr - User mailing list archive at Nabble.com.