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 AHMET ARSLAN <io...@yahoo.com> on 2009/11/08 12:11:17 UTC

schema.jsp is not displaying tint types corretly

I have a field defined tint with values 100,200,300 and -100 only.
When i use admin/schema.jsp i see 5 distinct values.

0	666083
100	431176
200	234907
256	33947
300	33947

First i thought that i post wrong values. I was expecting 4 distinct values.
When i query flag_value:256 i get 0 docs. Same as with flag_value:0.
100 200 and 300 works as expected. So i concluded that there is something wrong with schema.jsp with tint types. I am using solr-2009-11-03.tgz

By the way, trie types makes sorts faster? Or they are only useful in range queries?


      

Re: schema.jsp is not displaying tint types corretly

Posted by Chris Hostetter <ho...@fucit.org>.
: So we can say that it is normal to see weird vaules in schema.jsp for trie types. Thanks for the explanations.

it's normal to see weird values in schema.jsp for all types, trie, stemmed, etc...


-Hoss


Re: schema.jsp is not displaying tint types corretly

Posted by AHMET ARSLAN <io...@yahoo.com>.
 
> I'm not an expert on Trie fields, but you need to remember
> that schema.jsp 
> shows you the *indexed* values, and the whole point of
> TrieFields is to 
> create multiple indexed values at various levels of
> precision so that 
> range queries can be much faster.
> 
> : When i query flag_value:256 i get 0 docs. Same as with
> flag_value:0.
> 
> ...which is to be expected since you didn't index any docs
> with those 
> exact values.  If you facet on that field, you should
> see the 4 values you 
> expect (and no others) because the faceting code for
> TreiFields knows 
> about the special values, but schema.jsp just tells you
> exactly what's in 
> the index.

Yes tint is the default one comes with schema.xml. I queried *:* and faceted on that field, result is just like as you said:
<lst name="flag_value">
<int name="100">1132679</int>
<int name="200">207459</int>
<int name="300">27474</int>
</lst>

So we can say that it is normal to see weird vaules in schema.jsp for trie types. Thanks for the explanations.


      

Re: schema.jsp is not displaying tint types corretly

Posted by Chris Hostetter <ho...@fucit.org>.
: I have a field defined tint with values 100,200,300 and -100 only.

i assume you mean that tint is a solr.TrieIntField, probably with precisionStep="8" ?

: When i use admin/schema.jsp i see 5 distinct values.
	...
: First i thought that i post wrong values. I was expecting 4 distinct values.

I'm not an expert on Trie fields, but you need to remember that schema.jsp 
shows you the *indexed* values, and the whole point of TrieFields is to 
create multiple indexed values at various levels of precision so that 
range queries can be much faster.

: When i query flag_value:256 i get 0 docs. Same as with flag_value:0.

...which is to be expected since you didn't index any docs with those 
exact values.  If you facet on that field, you should see the 4 values you 
expect (and no others) because the faceting code for TreiFields knows 
about the special values, but schema.jsp just tells you exactly what's in 
the index.


-Hoss


Re: schema.jsp is not displaying tint types corretly

Posted by AHMET ARSLAN <io...@yahoo.com>.
> maybe you indexed some documents with
> value 256, but then deleted them?  try optimizing to
> get the terms removed.

I am running full-import with DIH. No deletions. And domain of this type is exactly 100,200,300 and -100. I am using this SQL query to fetch that field:

SELECT CASE WHEN ... THEN 300 WHEN ... THEN 200 WHEN ... THEN 100 ELSE -100 END AS flag_value FROM ...

I just optimized just to make sure. The same:

Distinct:  7

0 987692 
100 751304 
200 236388 
256 33830 
300 33830 

Interestingly it says there is 7 distinct values. When i try to see top 7 terms it always shows top 5. Also changes the value of top terms textbox to 5 again.



      

Re: schema.jsp is not displaying tint types corretly

Posted by Erik Hatcher <er...@gmail.com>.
maybe you indexed some documents with value 256, but then deleted  
them?  try optimizing to get the terms removed.

	Erik

On Nov 8, 2009, at 6:11 AM, AHMET ARSLAN wrote:

> I have a field defined tint with values 100,200,300 and -100 only.
> When i use admin/schema.jsp i see 5 distinct values.
>
> 0	666083
> 100	431176
> 200	234907
> 256	33947
> 300	33947
>
> First i thought that i post wrong values. I was expecting 4 distinct  
> values.
> When i query flag_value:256 i get 0 docs. Same as with flag_value:0.
> 100 200 and 300 works as expected. So i concluded that there is  
> something wrong with schema.jsp with tint types. I am using  
> solr-2009-11-03.tgz
>
> By the way, trie types makes sorts faster? Or they are only useful  
> in range queries?
>
>
>