Re: Fwd: Re: [spfbis] SPFBIS proposed charter

Brian,

Excellent point.  MUST defines the absolute minimum criteria for 
compliance with an RFC.  SHOULD often defines best practice additions 
which extend an RFC.

As the charter of this RFC bis, as I understand it, is not to add any 
functionality (required as MUST or otherwise) to the existing draft 
specification from which the bis shall become based, the argument over 
extension compliance for this process seems moot.

If there is a desire to add functionality to the specification as a 
requirement to implementation (a MUST case) then perhaps that is for a 
V3 or V4 SPF task force to hash out.

Alan M.

On 12/18/2011 10:25 AM, Brian G. Peterson wrote:
> On Sun, 2011-12-18 at 09:29 -0500, HECTOR SANTOS wrote:
>> Thats fine, and yes Extension are Extensions when mean OPTIONAL. So
>> anything new added to the RFC bis, MUST NOT be rammed it down people's
>> throat using odd readings of RFC 2119. In other words, people are
>> still compliant with SPF and SPF BIS while completely ignoring this
>> ILLEGAL CROSS BOUNDARY scope extension.  I don't want to see
>> statements like you made recently such as:
>>
>>          "Put another way, you can't claim compliance with this document
>>           unless you apply the SHOULD, but extant implementations are
>>           otherwise completely unaffected.
>>
>> IOW, you CAN NOT make this cross boundary scope extension a SHOULD for
>> SPFBIS if you truly believe what you stated.   If you feel
>> differently, then this not BIS work but a new protocol and new mandate
>> for a new version of SPF - a different protocol.
>
> I'm not taking sides here, since I am not really familiar with the
> technical details anymore, but a implementation is compliant if it meets
> all of the MUST directives.
>
> SHOULD directives have always been strong recommendations, but in no
> standards process that I have ever been associated with (and there have
> been many) have the *recommendations* in SHOULD directives been required
> for compliance. You may not have as broad of interoperability as you
> would like without implementing SHOULD directives, but that doesn't mean
> you are non-compliant.
>
> Now, I would suggest from plain meaning of the word 'Extension' that no
> extension would ever be required for standards compliance. MUST and
> SHOULD directives could still be applied within each extension method to
> make the *extensions* compliant and inter-operable with each other.
>
> Cheers,
>
>      - Brian
>