Trekking the Road from SPDY to HTTP/2 in Firefox | Hacker News Reader

Trekking the Road from SPDY to HTTP/2 in Firefox

bitsup.blogspot.com

One thing this doesn't seem to answer... Why bother when we already have SPDY?

Or HTTP over SCTP

As soon as Microsoft embeds SCTP in every Windows device, we might be able to start discussing this...

SCTP can be implemented in a shared library and tunneled through UDP. That actually works better since otherwise NATs kill you

Isn't the idea to have a formal standard owned by a standards body? Currently SPDY has not been accepted by any standards body as a formal standard.

That's probably why HTTP/2.0 is going to be based on SPDY, an attempt to codify an emerging standard.

Renaming SPDY to HTTP/2.0 is on the table, but some people disagree with some of SPDY's design choices and are hoping to fix them.

Because SPDY has well known flaws (let's call them design choices) for example in caching. There have been many discussions in HN about these topics.

Also, because there is no such a thing as "SPDY" what you are referring to is "SPDY 2 as implemented by Google Chrome and almost fully documented by Google". SPDY is a work in progress. If you want a real deployment to be possible, you need to have at least a fully documented basic core that has been discussed openly, not just an implementation from a corporation subject to changes whenever internal pressures require it.

SPDY is an IETF draft, working toward a ratified standard. https://tools.ietf.org/html/draft-mbelshe-httpbis-spdy-00

> SPDY is an IETF draft, working toward a ratified standard. https://tools.ietf.org/html/draft-mbelshe-httpbis-spdy-00

Now it is, before it was not. And that is exactly my point in my reply: you cannot take whatever you call "SPDY" at the moment (implementations + wiki pages) and call it "HTTP 2". You need to document it, propose it, discuss it openly, accept criticism, fixes and, if needed, radical changes. This is what is happening now in the HTTPbis working group and is a good thing, but it is not going to be a rubber-stamping process for SPDY as it is now, I hope.

SPDY patches some of the problems in the more common use cases for HTTP for today's applications. It doesn't really change much of the spec, and it doesn't even pretend to address potential future use cases. Here's a recent criticism by the lead architect of Varnish https://news.ycombinator.com/item?id=4253538

On the other hand, SPDY never tried to address those issues: the aim was to create a more efficient over-the-wire transport layer for HTTP/1.1 semantics.

I was answering this question https://news.ycombinator.com/item?id=4346116 Your point is exactly why we shouldn't settle for SPDY becoming HTTP 2.0.

Standardization is a critical step. It verifies that we use open IP which is available to everyone, provides a open forum for discussion, and archives the protocol in a way all vendors can use to resolve disputes about how we communicate.

Reply on news.ycombinator.com