automatic facet.limit?

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

automatic facet.limit?

Yonik Seeley-2
I was just testing the new faceted browsing with a large dataset (25M
records) and did
facet=true&facet.field=author

*boom*... firefox hangs hard after getting back megabytes and
megabytes of XML, and the appserver heap isn't too happy.

I put in a built-in default limit of 10 returned docs for just this
reason... to prevent people from shooting themselves in the foot when
learning/fooling around with Solr.  A person new to Solr might not
even realize *why* firefox just hung on them when they tried the
faceting feature.

Should we do the same for facet fields?  But that means we might need
a number to specify "no limit" (like 0 or -1).

-Yonik
Reply | Threaded
Open this post in threaded view
|

Re: automatic facet.limit?

Mike Klaas
On 9/12/06, Yonik Seeley <[hidden email]> wrote:
> I put in a built-in default limit of 10 returned docs for just this
> reason... to prevent people from shooting themselves in the foot when
> learning/fooling around with Solr.  A person new to Solr might not
> even realize *why* firefox just hung on them when they tried the
> faceting feature.
>
> Should we do the same for facet fields?  But that means we might need
> a number to specify "no limit" (like 0 or -1).

+1.  Solr should make some effort to behave reasonably unless
explicitly directed otherwise.

-Mike
Reply | Threaded
Open this post in threaded view
|

Re: automatic facet.limit?

Chris Hostetter-3

: > Should we do the same for facet fields?  But that means we might need
: > a number to specify "no limit" (like 0 or -1).
:
: +1.  Solr should make some effort to behave reasonably unless
: explicitly directed otherwise.

I suppose we should ... i avoided having a built in limit initially
because there is added work neccessary to limit the results (managing the
BoundedPriorityQuery) that i didn't want t do unless neccessary -- but you
guys are right, as long as there is a way to trigger the unbounded
behavior (with limit=0 or limit=-1) having a "sensible" upper bound by
default makes sense.

I'll take care of it.


-Hoss

Reply | Threaded
Open this post in threaded view
|

Re: automatic facet.limit?

Yonik Seeley-2
On 9/12/06, Chris Hostetter <[hidden email]> wrote:
> as long as there is a way to trigger the unbounded
> behavior (with limit=0 or limit=-1) having a "sensible" upper bound by
> default makes sense.

A limit of 0 can make sense, so mapping -1 to unbounded seems like a
better choice.

-Yonik