[hibernate-dev] hibernate-dev Digest, Vol 56, Issue 20

[Hardy Ferentschik]

On Mon, 28 Feb 2011 19:56:32 +0100, Marc Schipperheyn  
<m.schipperheyn at gmail.com> wrote:

> My two cents:
> * Most of the magic of Bobo Browse apart from the faceting itself is in  
> the performance. I think the challenge with any kind of faceting is  
> keeping it performant which Bobo achieves through all sorts of optimized  
> caches.
> * The problem for me with Bobo is real time updating the Bobo index which
> happens only through an add on that is not so easy, for me, to integrate
> with HibSearch.
>
> Where I spent most of my time customizing Bobo for my needs is:
> * label conversion: converting values to readable labels and back. E.g.  
> you store a date and want to display facets based on "yesterday" etc.
> * value conversion: converting a value to another value
> * Working with RunTime values, such as dates, vs normal pre determined
> values
> * debugging issues. The ability to be able to see what is happening under
> the hood is crucial because the index is much harder to access than a
> relational database
> * loading category specific facet values such as price ranges etc
>
> Facets I use are
> * Simple facets: just facet on a value
> * Location facets: geo locating based on rage in miles or kilometers
> * Path facets: drilling down through a facet, e.g. from Cities to
> City=>Neighbourhoods
> * Pre determined ranges, often filtered based on category specific values
> * Multi value facets, where you select based on property that can have
> multiple values

Thanks for this use-cases and integration reports. I will definitely take  
them
into consideration moving forward.

> Not sure if this helps but there it is. Bobo is a great product but it  
> took me a lot of time to get it "right"

That was my impression as well. It would take quite a long time to get it  
right
and make it fit into Search. For this reason I thought it would be best to  
start
with a easier implementation. The Collector code is actually quite simple  
and easy
to understand, especially after making some of the improvement Sanne  
mentioned.
Biggest concern on my behalf atm is memory footprint.

There was also just recently some talk on the Lucene side to integrate the  
Core Solr
faceting into Lucene (not the request, response parts, but the actual  
facet counting
impl). That would be a great feature and would obviously benefit us.  
Unfortunately,
Lucene is also not very fast with releases ;-)

--Hardy