- From: Amos Jeffries <squid3@treenet.co.nz>
- Date: Thu, 13 Jun 2019 19:38:20 +1200
- To: ietf-http-wg@w3.org
On 13/06/19 6:27 pm, Carsten Bormann wrote: > On Jun 12, 2019, at 19:26, Julian Reschke wrote: >> >>> 2) Assuming the answer to (1) is no, what should we make of RFC7030 >>> that says to use it, and to base64 binary objects? >> >> Raise an erratum :-). > > Not sure about the smiley here. This is an erratum. > Now how to resolve this depends on what option causes the smallest damage. > > RFC 7030 is supposed to be a widely deployed protocol, so it should not be hard to find out what people out there are actually doing. > > Grüße, Carsten > IME the desired behaviour is for the agent mapping into HTTP(S) to decide between three actions, in order of preference: 1) remove the encoding and send binary of the resources native Content-Type. 2) if all agents end-to-end are wanted to retain the same encoding - AND the final recipient is known able to decode; then C-T-E can be mapped to HTTP's Content-Encoding header and the body sent as-is. 3) if the next hop (only) is expected to be able to decode, but other hops uncertain; then C-T-E can be mapped to HTTP Transfer-Encoding header and the body sent as-is. * T-E is a mechanism without widespread support. So this option usually turns out to be not available at the HTTP next-hop. Amos
Received on Thursday, 13 June 2019 07:39:13 UTC