Thanks, Michael:
I make a simple API that when I want to make the updated index to be
functional, re-assign the Singleton instance of Searcher to be null,
and that seems to work.
I'll check the SearcherManager for further optimization.
Thanks for your trouble to review this problem and it really
Great! Thank you for bringing closure.
Mike McCandless
http://blog.mikemccandless.com
On Mon, Feb 20, 2017 at 3:41 AM, Chitra R wrote:
> Hey, I got it. Thank you so much.
>
> On Sat, Feb 18, 2017 at 5:33 PM, Michael McCandless
> wrote:
>>
>> You'll need to make your own buildFacetResults met
An IndexSearcher only searches the point-in-time view of your index as
of when it was opened.
So any changes to the index (newly indexed documents, or deleted
documents) will not be reflected until you refresh the underlying
IndexReader, to see the next point-in-time view.
The refresh is efficien
Hey, I got it. Thank you so much.
On Sat, Feb 18, 2017 at 5:33 PM, Michael McCandless <
luc...@mikemccandless.com> wrote:
> You'll need to make your own buildFacetResults method in your
> DrillSideways subclass, and inside there you compute the facet counts
> for each dim using the implementation
> PatriciaTrie. In particular building an FST with doShareSuffix = false is
> the fastest of any option,
If you don't share the suffix then you are building a kind of Patricia
trie... But suffix sharing is cheap and can give you a memory saving
(and resulting cache locality sometimes) that is non-
