Re: [apps-discuss] Feedback on draft-moonesamy-rfc2369bis-01 and draft-moonesamy-rfc2919bis-01

[Alessandro Vesely <[e-mail]>]

On 09/Jan/12 23:02, S Moonesamy wrote:
> At 02:54 09-01-2012, Alessandro Vesely wrote:
>> If it doesn't have to be a real DNS name, the reason to treat
>> ".invalid" as a special case becomes rather obscure.
> 
> It's to avoid localhost.com, etc.

Since you said one doesn't have to be the registrant of localhost.com
in order to use it as a list-id, it's still not clear.  Does the spec
mandate that list-id-namespace MUST be an existing domain name?

>> Since you are somewhat diluting the spec, you might as well have
>> merged it with 2369bis.  Why not?
> 
> 2369bis is more about list commands whereas 2919bis is about list
> identifiers.

Nevertheless, BCP 167 proposes List-Post as a candidate domain
identifier.  Accordingly, DKIM driven heuristics should use that field
rather than List-Id.

> Keeping the them as two separate documents makes it easier to
> compare changes.

It is better to rely on the purposely provided appendixes.

By keeping them separate, you turn this state of affairs into a
tradition.  I mean that more stiffness will be needed to merge them in
the future, assuming they will happen to be rewritten simultaneously
again.

> Some parts of DNS are about policy.
> 
>   "Several people expressed considerable scepticism about the
>    chances of success in response to domain name registry issues."

Yeah, it is curious that we consider number databases more solid in
that respect.  (Is it because we only have five?)  To ease the
transition to IPv6 we should use names instead.  For DNS, one can
certainly blame the ADMD for any bad settings, so DNS might be
considered more consistent than registries.

One side effect of merging 2919 into 2369 is to corroborate the idea
that List-Id is tailored for discussion lists.  You may have
alternative ideas to express the same concept without merging.  In any
case, since the possibility to use it as a policy element of bulk
commercial messaging is not fully supported, it may make sense to tell it.

> P.S. As all messages from this mailing list are DKIM signed, you can
> use some heuristics to detect misuse of the list-id.

BTW, isn't it worth to reference BCP 167?