3 ms·
I still feel like this is just being rammed through far far too fast. SPDY might well be the right way to go, but it's only been deployed for a small number of
by stormbrew 13y ago
I still feel like this is just being rammed through far far too fast. SPDY might well be the right way to go, but it's only been deployed for a small number of very large sites for a couple of years at this point. It was a foregone conclusion when the schedule was announced for http2.0 that it would be spdy, because only spdy has demonstrable value in production. There was no time for anyone to come up with or evaluate an alternative. But on the other hand, there hasn't been near enough time to demonstrate that spdy is good for the whole internet, which is what blessing it with the http protocol name implies.
And I just don't see why it should be rammed through. SPDY has a connection upgrade path, that afaik is largely unchanged for http2.0 anyways.
- drivebuy 13y ago> because only spdy has demonstrable value in production. What demonstrable value? Google has never compared it to pipelining and when Microsoft did so they found it about the same and sometimes slower. There is next to zero value in SPDY over pipelining across several connections. It's just complicated for no good reason.
- stormbrew 13y agoI think you're seeing a defence of spdy in my post that isn't really there. By value I mean simply that it is at least functional. Any alternative proposal, given the short proposal cycle, would not even have that.