#360 (advice on defining new cache directives), was: #307 (untangle Cache-Control ABNF)

On 2012-06-14 05:23, Mark Nottingham wrote:
> ...
> E.g., we don't place ANY conformance requirements on new headers; it's all advice: <https://svn.tools.ietf.org/svn/wg/httpbis/draft-ietf-httpbis/latest/p2-semantics.html#considerations.for.creating.header.fields>
> ...

And indeed we don't have this kind of advice for Cache-Control 
directives, but I think we should.

I just opened <http://trac.tools.ietf.org/wg/httpbis/trac/ticket/360>:

> For new cache control directives, we should recommend to document
>
>     what it means for a directive to be specified multiple times (invalid?)
>
>     what it means for an argument-less directive to have an argument
>
>     what it means for a argument-required directive not to have an argument

I believe we should improve the spec here no matter what we do about 
issue #307.

My 2 cents:

> For new cache control directives, we should recommend to document
>
>     what it means for a directive to be specified multiple times (invalid?)

I think the least surprising approach is to recommend to make this 
invalid. Mark noted that there may be cases out there where it's both 
valid and useful; feedback appreciated...

>     what it means for an argument-less directive to have an argument
>
>     what it means for a argument-required directive not to have an argument

A simple approach would be to recommend to make a missing argument to be 
equivalent to a zero-length string argument.

Re #307: optimally, these rules would be the same for all cache 
directive, or for at least all new directives. But if we can't do that, 
we should at least recommend a common approach.

Best regards, Julian

Received on Thursday, 14 June 2012 10:51:44 UTC