one way I have done it was to scope repo's and wha...

# rx

[trevjones]

one way I have done it was to scope repo's and what I called stream factories to a @UserSessionScope then the VM/C/P would use the stream factory. the repo would just have singles. the stream factories would use the singles inside an observable impl and poll the server for incremental updates. and finally the stream factories exposed observables that were ref counted so disposal was mostly magic and worked well if the consumers lifecycles overlaped (create/destroy usually does it well enough)