[QUOTE]"Once browsers and servers learn how and when to take advantage of that, performance should start incrementally improving."[/QUOTE]
Cool, can't wait to see some of those performace improvements at the end of the next decade.
Watch as it turns out to be shite.
god damn, when did http 1.1 come out anyways? it must've been a long-ass time ago.
[QUOTE=Jeep-Eep;47168022]Watch as it turns out to be shite.[/QUOTE]
It's really just HTTP 1.1 but binary.
The actual improvements are deferred to a later version to drive adoption of this one.
[QUOTE=Tamschi;47169420]It's really just HTTP 1.1 but binary.
The actual improvements are deferred to a later version to drive adoption of this one.[/QUOTE]
You're right that it's binary, but are you thinking of something else? The improvements are core to the protocol (It inherited multiplexing from SPDY, and priorities were added a while back and are already supported by clients)
[QUOTE=Im Crimson;47166815]Cool, can't wait to see some of those performace improvements at the end of the next decade.[/QUOTE]
Firefox 35 already supports it, and they're seeing ~9% of all traffic using it already.
[QUOTE=TheDecryptor;47169630]You're right that it's binary, but are you thinking of something else? The improvements are core to the protocol (It inherited multiplexing from SPDY, and priorities were added a while back and are already supported by clients)
[...][/QUOTE]
Oh right, I forgot about that part. It's still (almost) completely translatable into and from 1.1 though.
What's planned for the next version is cleaning up the headers etc. I hope that means they'll stop sending quite as much uniquely identifiable information.
The headers situation is interesting, it inherited them from HTTP 1, but handles them in a completely different manner that it's not really applicable (And everybody who implements HTTP2 treats it as an encoding of HTTP 1, so they re-use the existing header format for input and output). I think we'll see the headers split off into a separate specification actually, it doesn't make sense to talk about them in context of HTTP2.
One change I really would like to see is HTTP2 over SCTP or QUIC, they overcome the HOL blocking you get with TCP when you get packet loss, but they require tight linking between the on-wire format and the underlying transport (Tying a HTTP2 stream to a SCTP/QUIC stream). Not so useful if you're on a good connection, but for somebody on satellite internet it could be huge.
Sorry, you need to Log In to post a reply to this thread.