If I upgrade to using the edismax parser in my fq, I get the desired
results.
The default lucene parser on fq must not be able to parse the more complex
nested clauses
q=*:*&fq={!type=edismax}((-(field:V1) AND -(field:V2)) AND -(field:V3)) -
Works
On 4/22/15, 3:27 PM, "Dhutia, Devansh" <ddhu...@gannett.com> wrote:
>I don’t know if that’s completely true, or maybe I’m misunderstanding
>something.
>
>If it doesn’t support purely negative subqueries, this shouldn't work,
>but
>does:
>q=*:*&fq=(-(field:V1))
>
>However, for me, the following is a summary of what works & what doesn’t.
>q=*:*&fq=(-(field:V1)) - Works
>
>q=*:*&fq=((-(field:V1) AND -(field:V2)) AND -(field:V3)) - Doesn’t
>work
>q=*:*&fq=(-(field:V1) AND -(field:V2) AND -(field:V3)) - Works
>q=*:*&fq=((*:* -(field:V1) AND -(field:V2)) AND -(field:V3)) - Works
>
>
>
>
>On 4/22/15, 3:02 PM, "Jack Krupansky" <jack.krupan...@gmail.com> wrote:
>
>>A purely negative sub-query is not supported by Lucene - you need to have
>>at least one positive term, such as "*:*, at each level of sub-query.
>>Try:
>>
>>((*:* -(field:V1) AND -(field:V2)) AND -(field:V3))
>>
>>-- Jack Krupansky
>>
>>On Wed, Apr 22, 2015 at 10:56 AM, Dhutia, Devansh <ddhu...@gannett.com>
>>wrote:
>>
>>> I have an automated filter query builder that uses the SolrNet nuget
>>> package to build out boolean filters. I have a scenario where it is
>>> generating a fq in the following format:
>>>
>>> ((-(field:V1) AND -(field:V2)) AND -(field:V3))
>>> The filter looks legal to me (albeit with extra parentheses), but the
>>> above yields 0 total results, even though I know eligible data exists.
>>>
>>> If I manually re-write the above filter as
>>>
>>> (-(field:V1) AND -(field:V2) AND -(field:V3))
>>> I get the expected results.
>>>
>>> I realize the auto generated filter could be rewritten in a different
>>>way,
>>> but the question still remains, why is the first version not returning
>>>any
>>> results?
>>>
>>> Solr does not report any errors & returns successfully, just with 0
>>> results.
>>>
>>> Thanks
>>>
