Roster load storm on startup

[valeriansaliou]

When the Prose app starts, I receive multiple contactChanged and roomsChanged events, leading to local roster invalidation and reload. This ultimately leads to the app requesting the roster multiple times to the server in parallel.

• Is there any issue where I am receiving those delegate events from the core client more than I should?
• Or should this be handled in the application, eg. by debouncing the load or changing the roster reload mechanics, eg. not reloading on those events and completely de-correlating the roster from those events?
• Would that have any impact on rooms?

Those screenshots show the event/stanza storm on startup, which happen in a second or so:

[valeriansaliou]

Poke @nesium

[nesium]

I had that on my list for a while. I'll give it a go. Would like to handle it in the core.

[valeriansaliou]

Perfect, sounds good yes!

[valeriansaliou]

I think that ideally the roster should be loaded once by the core, and and then maintained w/ roster pushes from the server. Calling load roster on the app's side would not load the roster from the server if called a second time, rather return storage-backed roster data.

[nesium]

Events are now throttled and coalesced. Please note the slight change in the delegate signature:

Edit: Messed up the link in the commit message (https://github.com/prose-im/prose-core-client/commit/fb1f7f274cfe4c1a8cfcc6a362d4612caa0e7b95)