Ah I see, the detail I missed was that in HTTP/1.1 the Host header is not part of control data when using relative URLs. Thanks for clarifying. David On Wed, Jun 29, 2022 at 5:52 PM Martin Thomson <mt@lowentropy.net> wrote: > On Thu, Jun 30, 2022, at 05:42, David Schinazi wrote: > > <<An intermediary that forwards a request over HTTP/2 MUST construct an > > ":authority" pseudo-header field using the authority information from > > the control data of the original request, unless the original request's > > target URI does not contain authority information (in which case it MUST > NOT generate ":authority").>> > > Note the restoration of the final missing parenthetical. > > > Am I misunderstanding something? > > The distinction here is between requests that a client creates for itself > and requests that it is forwarding. > > The interpretation from Stefan is right: if you forward a request that > only has Host and not :authority (perhaps because it came in HTTP/1.1) then > you include "host" and not ":authority" in HTTP/2. > > Consider also that not every HTTP/2 client conforms to RFC 9113; there are > a number that conform to the language in RFC 7540. > > (If your next point is that this could be clearer in the spec, I'm not > going to disagree with you.) > >Received on Thursday, 30 June 2022 20:15:25 UTC
This archive was generated by hypermail 2.4.0 : Thursday, 2 February 2023 18:44:07 UTC