Re: [Gen-art] Genart Last Call review of draft-freed-smtp-limits-06

[John C Klensin <[e-mail]>]

Murray, Paul,

Please stand by for half-written note...   I should be able to
finish and post the next couple of hours.

  john


--On Sunday, September 17, 2023 14:53 -0700 "Murray S.
Kucherawy" <[e-mail]> wrote:

> On Fri, Sep 15, 2023 at 9:49 AM Paul Kyzivat
> <[e-mail]> wrote:
> 
>> 
>> Section 7.2 seems to conflate two things:
>> 
>> - the information that must be provided in a specification
>>    document that registers new limits
>> 
>> - the information that is to included in the registry itself
>> 
>> ISTM that the registry itself should contain the limit name
>> and a reference to the specification document. It might also
>> contain the value syntax, or at least an indication if a
>> value is allowed.
>> 
>> The descriptions of semantics, restrictions, and security
>> considerations don't lend themselves to inclusion in the
>> registry, but should be clearly spelled out in the
>> specification.
>> 
> 
> This is an interesting observation.  I suppose I overlooked it
> because by now I'm used to both types of registries.  A good
> example of this style is the media types registry, where some
> of the details (e.g., option names) that you might normally
> expect to find in the specification document only are actually
> also required to be present in the registry.  That model of
> registry has been around for a pretty long time and we seem to
> be fine with it.  But most other newer registries are just a
> table of the reserved name and a reference to the specifying
> document, with all of the details typically stored in the
> latter, with maybe a "status" column included.
> 
> I'd be inclined to split the difference, and say either the
> registry has to contain the limit value's syntax, or a
> reference to the defining document where such can be found.
> We allow this in the media types registry for security
> considerations, for example.
> 
> 
>> Also, the request for the new registry should probably
>> include its exact name ("SMTP Server Limits"?), and that it
>> should be included within the "MAIL Parameters" protocol
>> registry.
>> 
> 
> I agree, precision here is never a bad thing.
> 
> -MSK