Re: Adam Roach's Discuss on draft-ietf-httpbis-expect-ct-07: (with DISCUSS and COMMENT)

[Adam Roach <[e-mail]>]

On 9/13/18 6:23 AM, Julian Reschke wrote:
> On 9/13/2018 8:54 AM, Adam Roach wrote:
>> ... > §2.1:
>>
>>>   Expect-CT           = #expect-ct-directive
>>>   expect-ct-directive = directive-name [ "=" directive-value ]
>>>   directive-name      = token
>>>   directive-value     = token / quoted-string
>>
>> I note that there is no registry for directive names in the IANA 
>> section, so
>> presumably there is a small, closed set of directives allowed here. 
>> Typically,
>> when this is the case, the ABNF includes the permissible values; e.g.:
>>
>>     directive-name      = "report-uri" / "enforce" / "max-age"
>>
>> ...although I also note that list item (5) under the ABNF implies 
>> that the
>> intention here is to be extensible. If such is the case, I would suggest
>> adding an IANA registry that records Expect-CT directives, and 
>> specifying the
>> ABNF as:
>>
>>     directive-name      = "report-uri" / "enforce" / "max-age" / token
>> ...
>
> Disagreed. We have stopped doing this in the HTTP specs for a reason - 
> it conflates two different thing: parsing, and detecting certain 
> predefined tokens. 


Thanks. If HTTP-related RFCs do this consistently, then I agree that you 
should keep with that convention.

/a