Re: BNF for Cache-Control

Adrien de Croy wrote:
> I guess the answer lies in the wording around the directives 
> themselves.  Some of them it's pretty clear are intended only for 
> request messages, some for response messages.  Some are less clear, and 
> the syntax for some differs between request and response (e.g no-cache).

In case you didn't notice: those have been separated into distinct 
subsections since draft 05.

> So one way to look at this would be that the grouping of directives into 
> cache-request-directive and cache-response-directive are mainly to 
> demonstrate which type of messages these directives are expected to be 
> used in, without banning them from the other.
> 
> I guess a similar issue would arise in any header that could be part of 
> a request or response message which contains directives which may or may 
> not make sense for a request or a response (e.g. Connect).
> 
> In the end, I'm more than happy to ignore headers that don't make sense 
> in the context, but specific wording in the spec to back those decisions 
> up would be useful.

The descriptions already should clearly say "general", "request", 
"response", or "entity"...

Do you think that is not sufficient? Maybe you could clarify, based on a 
concrete example?

BR, Julian