Support bServerSide for datatables

[jean]

When viewing a large view, a dynamic table becomes impractical (e.g. will freeze Chrome for a couple of minutes).

For this, views will need to expose an URL that returns blocks of JSON for the requested rows.

See http://datatables.net/usage/server-side

[ebrehault]

True.
I will try that.
I guess we will have to adapt the live filtering mechanism (we cannot ask Plomino to return all the docs containing the letter 'a' if there are a lot of docs).

[ebrehault]

ok done
I have tested with a view containing 45000 docs, it just rocks !

Note: filtering only works if local fulltext index is available, but if it is not (and you do not want to enable it), you may set bServerSide: false manually in the datatables settings.

[jean]

Cool :)

[jean]

Currently every "Next" page will result in querying the catalog for all records that match the view, and creating a Batch instance with all records. So if a view has 10 pages, this will mean 10x more CPU used, and 10x claiming and freeing memory for search results and the Batch.

To mitigate this, a view that is being paged serverside should have a cache which has a tojson method which supports the bServerSide parameters for paging.

Since view results are the same for all viewers, there can be one cache which is invalidated when the view is updated.