Re: [http-state] draft-ietf-httpstate-cookie-05 posted

[Adam Barth <[e-mail]>]

On Sun, Mar 14, 2010 at 4:38 PM, Yngve Nysaeter Pettersen
<[e-mail]> wrote:
> On Thu, 11 Mar 2010 18:42:51 +0100, Adam Barth <[e-mail]> wrote:
>> On Thu, Mar 11, 2010 at 9:06 AM, Yngve Nysaeter Pettersen
>> <[e-mail]> wrote:
>>>
>>> On Thu, 11 Mar 2010 03:23:03 +0100, Adam Barth <[e-mail]>
>>> wrote:
>>>>>
>>>>> * 4.1 definition of Max-age is missing. It is used in 5.2.1
>>>>
>>>> This is intentional.  The folks in the working group didn't think we
>>>> should recommend Max-Age to server operators because it's not
>>>> implemented in Internet Explorer.  I'm optimistic that IE will
>>>> implement it, but that's why it's not there.
>>>
>>> I still think it should be there, perhaps with a note that not all
>>> clients
>>> support it (Opera didn't until a couple of years ago), and that servers
>>> should complement it with an Expires attribute.
>>
>> Feel free to convince the working group on this point.  I support
>> describing it in Section 4.  It seems odd to leave out only one of the
>> attributes.
>
> IMO the syntax section of a specification should be complete enough that a
> reader can know "at a glance" what elements the possibilities the
> specification embodies, even if there are agents that are currently not
> using some of the syntax elements. There should be no "surprise" attributes
> that show up later in the document, whose existence cannot be derived from
> being part of some kind of syntax group.
>
> If, as in this case, a particular element is not in use by at least one
> major client (plus older legacy clients), then that should be documented by
> at least a comment to the syntax, and other parts of the document should
> elaborate on that and specify how a server should use the attribute in an
> interoperable fashion.
>
> There may be situations, such as in HTTP, where the exact details for a
> given header is specified in a section (or multiple sections) of its own,
> but even then the header is mentioned in the syntax summary.
>
> While I would prefer the entire Max-age syntax in Section 4, one possible
> way to address the concerns mentioned could be to include a
> "max-age-attribute" element in Section 4, and then have the entire
> definition in the later section. IMO that would also look odd, and it is
> better to have the entire syntax in Section 4.

I think you misunderstand the structure of the document.  Section 4 is
not about the syntax of the protocol.  It's about a subset of the
protocol that servers can use reliably.  I need to make that clearer
in the document.  The syntax of the protocol is defined in Section 5.

Adam