Re: [hybi] [dispatch] Publication request for draft-bormann-hybi-ws-wk-00

[Anne van Kesteren <[e-mail]>]

On Thu, Jun 22, 2017 at 6:33 PM, Alexey Melnikov
<[e-mail]> wrote:
> On Thu, Jun 22, 2017, at 05:29 PM, Anne van Kesteren wrote:
>> On Wed, Jun 21, 2017 at 5:48 PM, Alexey Melnikov
>> <[e-mail]> wrote:
>> > RFC 5785 defined a path prefix, "/.well-known/", that can be used by
>> > well-known URIs, specifically for the "http" and "https" URI schemes.
>> > Other URI schemes like "coap" also opted-in into using .well-known.
>> >
>> > RFC 6455 defined WebSocket protocol and "ws"/"wss" URI schemes. It
>> > hasn't however opted-in into using the .well-known registry defined by
>> > RFC 5785. draft-bormann-hybi-ws-wk-00 is fixing this deficiency.
>>
>> In practice (and in the specification that supersedes parts of the
>> RFC) the scheme in use is already http/https:
>>
>>   https://fetch.spec.whatwg.org/#websocket-protocol
>
> I think this is an orthogonal thing, as ws/wss URI schemes are also used
> in not HTTP contexts. So basically draft-bormann-hybi-ws-wk-00 doesn't
> affect Fetch.

The /.well-known/ thing only makes sense with respect to the WebSocket
handshake, no? And that's already defined to be an HTTP request. So
I'm not sure why any /.well-known/ things with regard to HTTP would
not already apply. (One of the reasons Fetch made this
change/clarification is to make sure HSTS, CSP, etc. would all apply
to WebSocket as well.)


-- 
https://annevankesteren.nl/