On 2 September 2014 23:48, Ilari Liusvaara <ilari.liusvaara@elisanet.fi> wrote: > This removes most HTTP-specific stuff from the framing layer (except > for SETTINGS space and error codes space)[8]. > Ilari, while I do not agree with absolutely everything in your breakdown, I definitely agree with the basic approach you are setting. The fundamental issue I see here is that there is no separation between HTTP semantics and the framing layer. I believe this is the meta-issue that we should debate and decide if we wish to solve it or not. Until we have decided that fundamental question, I think we are bound to repeat many of the technical debates we have had without a resolution that satisfies a significant majority. This working group started with a choice between two starting points: speed and mobility (HTTP semantics over websocket framing) and SPDY (HTTP semantics over a new multiplexed framing layer). Together with the charter to consult websocket, it looked like it was pretty clear that the original intend was to support multiple semantics. But somewhere along the way, this clear separation was dropped and I believe most of the current disagreements stem from that decision. Rather than continue to debate "this is a poor design for a framing layer" vs "it works perfectly well for HTTP", I think we need to step back and have a meta discussion. Is it an issue or not that the framing layer is HTTP specific? regards -- Greg Wilkins <gregw@intalio.com> http://eclipse.org/jetty HTTP, SPDY, Websocket server and client that scales http://www.webtide.com advice and support for jetty and cometd.
Received on Tuesday, 2 September 2014 23:09:05 UTC