lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Zheng Lin Edwin Yeo <edwinye...@gmail.com>
Subject Re: Search through all fields in collection
Date Thu, 25 May 2017 13:12:32 GMT
Hi Shawn,

Yes, I couldn't use more than one field for the df parameter.

What I am worried about is the increase of the size of the index, and
affecting overall performance when using copyField, which is why we are not
very keen on using that.

When using dynamic fields for copyField, is it true that we have to copy
all or none of the same type. Meaning if my string is declared as *_s, when
I do a copyField, I have to copy all the fields with source = *_s, and
cannot select which individual fields with _s to copy?

Regards,
Edwin


On 25 May 2017 at 21:00, Shawn Heisey <apache@elyograg.org> wrote:

> On 5/25/2017 12:45 AM, Zheng Lin Edwin Yeo wrote:
> > Would like to check, is it possible to set a configuration in
> > solrconfig.xml whereby the search will go through all the fields in the
> > collections?
> >
> > Currently, I am defining the fields to be search under the "df" setting,
> > but unlike "fl", I could not set it to the value "*". The only way which
> I can think of currently is to list all the fields under the "df" setting.
> However, this doesn't seems like a very good way, so I'm checking if there
> are better methods.
>
> As far as I'm aware, the df parameter (default field) can only take one
> field.  I have never seen any ability to use more than one field there.
> If that actually works, then it is functionality that I've never seen.
>
> If you switch to the dismax or edismax query parser, then you can list
> multiple fields in the qf, pf, pf2, and pf3 parameters, but even then,
> Solr does not provide a simple way to specify all fields.  You can
> specify them all, but you have to do so one by one.
>
> Searching all fields is usually very expensive anyway.
>
> The method that Rick gave you, of copying all fields into a single field
> that you can specify as the default field, is one way to do it, but be
> aware that if you choose this method, every term will be analyzed in
> exactly the same way.  This may not be what you want.  It also can
> greatly increase the size of the index, affecting overall performance.
>
> Thanks,
> Shawn
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message