Re: [core] Stateless proxy and Request-Tag

[Christian Amsüss <[e-mail]>]

On Thu, Oct 31, 2019 at 05:59:11PM +0100, Carsten Bormann wrote:
> I don’t understand how Request-Tag became necessary for Block2.
> For me that is a Block1-only addition.

It is intended for Block1, but spills into the Block2 phase due to
RFC7959 Section 2.7's description of the block-2-phase requests as
"similar to the requests in the Block1 phase [without] the Block1
Options and [including] a Block2 request option with non-zero NUM".

The Block1 operation may have created state on the server, and the later
requests may be only accessible if the block-matching-relevant options
of the Block2-phase requests are the same as the earlier ones.

From that, I'd assume the use of Request-Tag to be suitable there.

How that relates to single-message unsafe requests with blockwise
response bodies (eg. empty POSTs) I'm unsure, but a stateless proxy
would send a Request-Tag on some of the messages as it can't tell them
apart from the above case.

KR
c

-- 
To use raw power is to make yourself infinitely vulnerable to greater powers.
  -- Bene Gesserit axiom