Collection Search - Omoccurrences Performance Optimizations

[MuchQuak]

Optimizations

Remove Count queries From base Search

General Problem is getting the search count is more expensive than doing the search and generally by a pretty significant margin.

Options

1. If pagination as it currently is must not change. Then only compute the value on initial load an only do page searches on subsequent loads.

2. Move to only previous and next for pagination. This would remove the need to the count query at all which would be a large performance boost. It would be a feature degradation and that is a trade off we will need to talk to uses about. Enhancement of this allow for page navs to previously visited pages so if all pages are visited it would appear no different (Unlikely that a user paginates through all the results).

3. Delay the loading of the count query. Basically allow the user to request or lazy load a count and page by page navigation system after searches has loaded. Pros are users who don't care about the exact count or exact page nav would see massive search improvements and cons are that users who do care will need to ask and wait for the info. With this method it would still be recommended to only load this value once.

Other notes

Some of these solutions could be paired together for a better overall experience.

[themerekat]

I think people do want counts when they run queries, though