>
> Why not a MUST? If you don't implement preauthorization, there's no need
> for a value.  I guess you could provide "null", but ... don't.


I have no objections to using MUST over SHOULD. Updated in 5fcaa6c on my PR
branch.

- Daniel / cpu

On Fri, Jan 5, 2018 at 4:24 PM, Richard Barnes <[email protected]> wrote:

> Why not a MUST? If you don't implement preauthorization, there's no need
> for a value.  I guess you could provide "null", but ... don't.
>
> On Fri, Jan 5, 2018 at 2:33 PM, Daniel McCarney <[email protected]>
> wrote:
>
>> Hi Ilari,
>>
>>
>>> This impiles that if pre-authorization is not supported, newAuthz better
>>> be absent (this is also useful for detecting pre-authorization support if
>>> the client can use it somehow).
>>> I think this should be at least a SHOULD.
>>
>>
>> You're right. My intention is to make the presence of that field an
>> indicator of pre-authorization support (or lack thereof). I'll update to a
>> SHOULD in my PR since I agree that maps better.
>>
>> Thanks!
>>
>>
>>
>> On Fri, Jan 5, 2018 at 2:27 PM, Ilari Liusvaara <[email protected]
>> > wrote:
>>
>>> On Fri, Jan 05, 2018 at 02:05:22PM -0500, Daniel McCarney wrote:
>>> > In Section 7.4.1 "Pre-Authorization" the spec says:
>>> >
>>> >  If a CA wishes to allow pre-authorization within ACME, it can offer
>>> > >  a "new authorization" resource in its directory by adding the field
>>> > >  "newAuthz" with a URL for the new authorization resource.
>>> >
>>> >
>>> > That text indicates that the CA may wish to *not* support
>>> pre-authorization
>>> > in which case the "newAuthz" resource will not be present in the
>>> directory.
>>> > E.g. the Let's Encrypt ACME v2 directory[0] does not include this
>>> resource.
>>> >
>>> > I think this should be further emphasized in Section 7.1.1 "Directory"
>>> > since this is where a developer is likely to look first when trying to
>>> > determine what directory fields can be assumed present.
>>> >
>>> > I opened a PR to add a small note about this to Section 7.1.1:
>>> > https://github.com/ietf-wg-acme/acme/pull/384
>>> >
>>> > This is perhaps a step past the line of being strictly editorial. I
>>> would
>>> > appreciate if anyone with objections raise them on-thread over the
>>> next few
>>> > days.
>>>
>>> Why MAY? If implementation does not support some functionality that an
>>> endpoint is exclusively for, it better omit that endpoint. This impiles
>>> that if pre-authorization is not supported, newAuthz better be absent
>>> (this is also useful for detecting pre-authorization support if the
>>> client can use it somehow).
>>>
>>> I think this should be at least a SHOULD.
>>>
>>>
>>>
>>> -Ilari
>>>
>>
>>
>> _______________________________________________
>> Acme mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/acme
>>
>>
>
_______________________________________________
Acme mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/acme

Reply via email to