Re: issue 381: Discovery of the support of the HTTP2 protocol: DNS-based Upgrade

On 20 February 2014 01:23, Mark Nottingham <mnot@mnot.net> wrote:
> This is what’s concerned me in the past. We’ve made a lot of decisions based upon the assumption that introducing h3…hn (as well as other protocol identifiers) won’t require adding more round trips.

In the general case, I don't think that this is a concern.  There is
an expectation that somewhere you are able to select between nothing
(i.e. HTTP/1.1) and "h2".  As long as that is general enough,
selecting "h3" instead should be no concern.  ALPN does that
trivially.

That doesn't mean that it can't be screwed up somehow.  Anything that
was serialized somehow (nothing -> h2 -> h3 -> h4 ...) would be an
issue, but then that's also an issue for HTTP/2, because it puts
HTTP/2 at a disadvantage.

I'm confident that, once we are aware of this constraint (which may
happen in time), we won't allow a solution that doesn't have such
undesirable properties.

Received on Thursday, 20 February 2014 17:15:43 UTC