This guy has some good rants on websockets, saving me the trouble. I’m strongly annoyed by the deficiencies of some aspects of the spec, which forces framing logic into the application layer unnecessarily and expensively, given that it’s stupidly mandated that the underlying protocol is TCP. I find this stuff very frustrating, and think that when we publish a recommendation for “RFB over the WebSockets transport” we’ll just throw out framing altogether. Our RealVNC server and websockify.js ignore the underlying framing and add in a cheap frame header on every flush, which I think is going to have to be the way to go for serving protocols expecting an underlying stream.