Re: [Last-Call] Last call feedback on draft-ietf-ohai-ohttp-05

[Martin Thomson <[e-mail]>]

Thanks for the concrete suggestions for text Mark, this is really helpful.  I've taken those here: https://github.com/ietf-wg-ohai/oblivious-http/pull/223

Responses to questions/issues inline.

On Wed, Nov 30, 2022, at 14:38, Mark Nottingham wrote:
> * The term 'Obvivious Relay Resource' is a bit odd. Section 6.2 admits
> that it can be a normal HTTP intermediary (i.e. a proxy) as long as a
> few constraints are observed. So, it doesn't need to be a resource.
> I know that's the fashion with the MASQUE folks, but I'd suggest that
> this just be called an 'Oblivious Relay'  -- or, indeed, 'Oblivious Proxy'.

Yes, it doesn't need to be a resource, but I think - in the spirit of the MASQUE work, which I happen to think is taking a positive direction - a resource is appropriate here.  That is, there is a resource that the relay operates with an identity of its own.

I realize that the classic HTTP proxy scenario has the proxy receive something like:

POST https://gateway.example/some/resource HTTP/1.1
Header: fields
go: here

This doesn't give the proxy any control over how resource identity affects its functioning as a proxy.  For a proxy that operates exclusively as a proxy, that's probably not a big deal, but we're seeing MASQUE functions being added to servers that do other things.

> HTTP has no concept of a 'Target Resource'; it's just a 'resource' (see 
> 9110 3.1). An easy fix would just be to use lowercase 'target'. I'd 
> also consider lowercasing 'resource' throughout.

Hmm, fair call.  I started on this then realized that maybe we were Making Names Stand Out a little too much and followed that to its logical conclusion here: https://github.com/ietf-wg-ohai/oblivious-http/pull/224