3 ms·
Why HTTP/2.0 does not seem interesting (2012)
- bhouston 12y agoI'm of the opinion that nothing will be perfect and that you can not please everyone. That said, it is good to hear dissenting voices.
- synctext 12y agoYou indeed cannot please everyone. But not addressing complexity, simply adding more.. Not solving the cookies issue, it's not sufficient. Protocols are more then a spec, they express the norms&values of an ecosystem. Is privacy really appropriately valued?
- personZ 12y agoIs there some alternative specification that solves these "issues"? Or is this one of those hand-wavy "solve it" kinds of things, where we have the luxury of not considering or worrying about the inevitable downsides of alternative approaches?
- mantrax5 12y agoAs the saying goes, the best way to piss off everyone is to try to please everyone. It seems like W3C are trying to please everyone. Hence the result.
- zimbatm 12y agoIn regards to the Host header. One thing that always seemed awkward to me is that the request line can contain the full URI on proxies but not for end-points. I suppose for backward-compatibility. Instead of: "GET /some/path HTTP/1.1\r\nHost: example.com" We could have: "GET http://example.com/some/path http://example.com/some/path HTTP/2.0\r\n" https://tools.ietf.org/html/rfc2616#section-5.1 https://tools.ietf.org/html/rfc2616#section-5.1
- nly 12y agoIf you merged the transports (for example, everything was over HTTPS), we could just have "GET example.com/some/path HTTP/2.0\r\n"
- zimbatm 12y agoYes, and then add SCTP (over UDP) and we have most of SPDY's benefits :) As they say, if the protocol's not supported, HTTP 1.1 fallback is always supported.
- dang 12y agohttps://news.ycombinator.com/item?id=6014976 https://news.ycombinator.com/item?id=6014976