Proposal for #486: Requiring proxies to process warn-date

<http://trac.tools.ietf.org/wg/httpbis/trac/ticket/486>

We discussed this issue in Berlin briefly, and I had an action to make a concrete proposal.

In <https://svn.tools.ietf.org/svn/wg/httpbis/draft-ietf-httpbis/latest/p6-cache.html#header.warning>:

OLD:
---8<---
   If an implementation sends a message with one or more Warning header fields
   to a receiver whose version is HTTP/1.0 or lower, then the sender MUST
   include in each warning-value a warn-date that matches the
   Date header field in the message.

   If a system receives a message with a warning-value that includes a
   warn-date, and that warn-date is different from the Date
   value in the response, then that warning-value MUST be deleted from the
   message before storing, forwarding, or using it. (preventing the
   consequences of naive caching of Warning header fields.) If all of the
   warning-values are deleted for this reason, the Warning header field MUST
   be deleted as well.
--->8---

NEW:
---8<---
   RFC2616 made the Warning header field's warn-date component optional; it 
   was only required to be sent when the recipient's version was HTTP/1.0 or lower.

   However, deployment experience has shown that many (if not most) intermediaries
   do not process the Warning header as required by RFC2616. This results in 
   situations where the field can appear in messages where it is not applicable, because
   a warning-value has not been removed by an intermediary.

   As a result, this specification shifts responsibility for processing of Warning from 
   intermediaries to the recipient that is actually consuming them.

   Generators of Warning header fields MUST include in every warning-value
   a warn-date that matches the Date header field in the message. Recipients that
   process a Warning header field MUST ignore (and MAY remove before forwarding)
   a warning-value whose warn-date is different from the Date value in the response. 
--->8---

We'd also remove the brackets in the ABNF that denote that warn-date is optional.

Comments?

--
Mark Nottingham   http://www.mnot.net/

Received on Monday, 2 September 2013 00:48:24 UTC