Re: ExtendedDisMax Field Alias Question

[Jamie Johnson <je...@gmail.com>]

It's been some time since I've thought about this, but I wanted anyone
interested to know I created the following JIRA asking for this
feature.

https://issues.apache.org/jira/browse/SOLR-3598

On Sat, May 26, 2012 at 9:28 PM, Jamie Johnson <je...@gmail.com> wrote:
> Yeah cycles in general I agree are bad, but perhaps an option to also
> include the original field or special handling of the aliased field to
> support this.
>
> On Sat, May 26, 2012 at 3:26 PM, Jack Krupansky <ja...@basetechnology.com> wrote:
>> That would create an alias "loop", which is not supported.
>>
>> For example,
>>
>> http://localhost:8983/solr/select/?debugQuery=true&defType=edismax&f.person_first_name.qf=genre_s&f.person_last_name.qf=id&f.name.qf=name+person_first_name+person_last_name&q=name:smith
>>
>> in Solr 3.6 generates a 400 response status code with this exception:
>>
>> org.apache.lucene.queryParser.ParseException: Cannot parse 'name:smith ':
>> Field aliases lead to a cycle
>>
>> Maybe what you would like is an enhancement to permit an explicit
>> refererence to the underlying field rather than the alias in an alias
>> definition, like:
>>
>> &f.name.qf=field.name+person_first_name+person_last_name
>>
>> -- Jack Krupansky
>>
>> -----Original Message----- From: Jamie Johnson
>> Sent: Friday, May 25, 2012 8:37 PM
>> To: solr-user@lucene.apache.org
>> Subject: ExtendedDisMax Field Alias Question
>>
>>
>> I was wondering if someone could explain if the following is supported
>> with the current EDisMax Field Aliasing.
>>
>> I have a field like person_name which exists in solr, we also have 2
>> other fields named person_first_name and person_last_name.  I would
>> like to allow queries for person_name to be aliased as person_name,
>> person_first_name and person_last_name.  Is this allowed or does the
>> alias need to not appear in the list of fields to be aliased to (I
>> remember seeing something about aliases to other aliases is allowed)?
>> I could obviously create a purely virtual field which aliases all 3
>> but it would be nice if the parser could support this case.