Re: http://tools.ietf.org/html/draft-nottingham-httpbis-alt-svc as a normative reference in http/2

On Wed, Mar 19, 2014 at 4:18 PM, William Chan (陈智昌)
<willchan@chromium.org>wrote:

> On Wed, Mar 19, 2014 at 1:15 PM, William Chan (陈智昌) <willchan@chromium.org
> > wrote:
>
>> On Wed, Mar 19, 2014 at 1:09 PM, Patrick McManus <pmcmanus@mozilla.com>wrote:
>>
>>
>>>
>>> More pertinent of course is the fact that this is one plan for
>>> bootstrapping http/2 - and if its going to be used, like upgrade which I
>>> consider far messier, it should be documented to get consistency among
>>> those parties that wish to use it.
>>>
>>
> Oops, I missed responding to this part. My understanding is that Rob was
> the main person looking to upgrading from HTTP/1.1 to HTTP/2 in cleartext,
> and that they prefer HTTP Upgrade. So if you're including this for his
> benefit, I think that may have been a mistake. But I should let Rob speak
> for himself.
>
>

I wasn't trying to suggest this should subsume upgrade for Rob's case - I
was trying, perhaps unclearly, to highlight the analogy between alt-svc and
upgrade and for that matter ALPN too. Upgrade is one method for
bootstrapping http/2 that we expect will be used in the wild and (even
though you and I don't intend to implement it). It does, imo. make sense to
document it so that those interested can have an interoperable definition.

The same can be said of ALPN and Alt-Svc.

I don't think you'll find the next alt-svc document particularly
threatening to the schedule - the header is just a different serialization
of the frame and importantly it also includes a necessary error code which
is relevant to resolving issues when following the frame as well as the
header.