We need to explore how the new virtual list: https://github.com/etiennesegonzac/dom-scheduler Can use a pregenerated model vial cached content in the service worker (old way of using render cache)

POC with the virtual list and caching via sw: https://github.com/arcturus/dom-scheduler/tree/contacts-sw-cache The way we use the render cache here is that we create a /model.json file that contains the first chunk to be render and the number of contacts. That's the information that the list needs to do a quick first paint. Will try another approach that is to save the information in the own index.html

Interesting results ahead: - First paint packaged ~ 960 ms - First paint with SW ~ 1100 ms (The SW serving is increasing the time) - First paint with SW and Cache ~ 1000 ms (We save a bit but not much) Will need to explore if the code added to cache, or the need of an extra request is what is not making us have a huge gain in the performance.

Closing this as fixed, since several tests have been done. https://github.com/arcturus/dom-scheduler/tree/contacts-test https://github.com/arcturus/dom-scheduler/tree/contacts-sw-cache We need to check now the performance for SW and raw cache.