one of the weird side effects of building a resilient abstraction for websockets - it’s making me want to use it for all data fetching?? retries, buffering, refreshing sessions, just a bunch of nice features, and it’s crazy fast, so why not use it for all data transfer?
Conversation
I could shift all the data fetching complexity (graphql or whatever you’re using) to a partykit room and simply connect to it, and simultaneously make everything faster but reduce bundle size??
1
1
23
oh you want to use a browser cache? throw indexed db at it and get granular control instead of futzing with headers and browser quirks etc
of course this might not be for everybody, but gosh it feels like I got a glimpse of something special. must investigate after I ship v0.1
7
33
Replying to
have you seen vlcn.io & riffle.systems? Treat each client as node of distributed db. All reads/writes are local and framework syncs. This all seems really promising.