Re: New draft: Maintaining HTTP Integrity Fields in Cached Responses

["Roy T. Fielding" <[e-mail]>]

> On Aug 21, 2026, at 6:09 AM, Antun Jurkovic <[e-mail]> wrote:
> 
> Hi all,
> 
> I have submitted an individual Internet-Draft proposing focused
> maintenance to RFC 9111:
> 
> Maintaining HTTP Integrity Fields in Cached and Cache-Generated Responses
> draft-jurkovikj-http-integrity-cache-00
> https://datatracker.ietf.org/doc/draft-jurkovikj-http-integrity-cache/
> 
> RFC 9111 defines general rules for updating stored response fields,
> combining partial responses, and constructing responses from cached
> state. Content-Digest and Repr-Digest, defined by RFC 9530, and
> Unencoded-Digest, currently defined by
> draft-ietf-httpbis-unencoded-digest, apply to specific message content
> or representation data.
> 
> Applying those general cache rules without accounting for the
> semantics of these Integrity fields can associate a digest value with
> data different from the data over which it was calculated.

That's unfortunate for the design of those fields. Maybe the field definitions
should take that into account.

Placing new HTTP requirements on all cache implementations is not an option
unless all implementations already support those requirements.

Defining new requirements for the implementations of an extension header field
only need support by those implementing the header field, but they have to do so
within the general context of HTTP: they will pass through systems and caches
that do not implement those fields. A design that doesn't support that context
will not succeed in deployment, no matter what RFC is updated.

Cheers,

....Roy