Re: 103 (Early Hints) vs. response headers

[Willy Tarreau <[e-mail]>]

On Wed, Mar 15, 2017 at 12:02:53PM +0900, Kazuho Oku wrote:
> 2017-02-24 9:12 GMT+09:00 Mark Nottingham <[e-mail]>:
> > My .02 -
> >
> >> On 24 Feb 2017, at 2:27 am, Vasiliy Faronov <[e-mail]> wrote:
> >>
> >>    HTTP/1.1 103 Early Hints
> >>    Link: </another-resource>; rel=preload
> >>    Warning: 299 - "something is not quite right"
> >>
> >>    HTTP/1.1 200 OK
> >>    Date: Thu, 23 Feb 2017 16:49:43 GMT
> >>    Content-Type: text/html
> >>    Link: </another-resource>; rel=preload
> >>    Connection: close
> >>
> >>    ...text goes here...
> >>
> >> Should it log/display the warning (as applied to the 103 response), or
> >> discard it (as missing from the 200 response)?
> >>
> >> Should the spec for 103 be more explicit about this?
> >
> > My reading is that "officially", the Warning is not in the response; the server thought something was wrong early in the process, but then realised it was fine.
> >
> > So, it MAY log/display the warning, but if it doesn't, it's still conformant.
> >
> > Some more examples might help.
> 
> RFC 6265 states that a user agent "MAY ignore Set-Cookie headers
> contained in responses with 100-level status codes".

Yes but it's mostly as a warning for server side to know that any cookie
sent there may be ignored (since 1xx may be appear multiple times and be
silently skipped).

> So to me it seems that if we state in Early Hints that the headers of
> a 103 response is ones that are applied (speculatively) to the final
> response but not the informational response itself, then we'd be
> overriding RFC 6265.

I'm not seeing it this way. In fact you may decide to put some headers
there for this exact reason : while 1xx MAY be ignored, those implementing
103 MAY/WILL consider them. And you're sending 103 hoping that someone
will make good use of it, not as a guarantee, so I don't think it
contradicts 6265.

Cheers,
Willy