Re: co-chair judgment of consensus related to last call period of 23-Aug-2004 to 10-Sept-2004

[Daniel Quinlan <quinlan@xxxxxxxxxxxx>]

Dave Crocker <dhc@xxxxxxxxxxxx> writes:

> My concern is not about efficiency. It is about the fundamental
> ability to get any successful interoperability.

I'm concerned about that as well, but efficiency is a concern if you're
trying to process mail as quickly as possible.  The limit of 10
redirects/includes is a bit worrisome too.

> If there is no core requirement for support of a particular record
> type, then there will be clients that support only one record type and
> servers that support only another.  This means that the clients will
> not be able to access the information.
>
> The instant you talk about "duplicate queries" you are assuming that
> the client supports more than one type.  The question is whether they
> are required to.  That is, do they have a MUST for each type?

There definitely needs to be a shared MUST for one type.  At that point,
though, how much point is there in even mentioning the other?

I absolutely don't want to see a MUST for two types for the receiver if
there's no schedule for phasing TXT out.

> For Internet-scale services there can be no such thing as a flag day.
> There is not central authority to assert it.  Even if there were,
> there is no way to coordinate that scale of operation.  Even centrally
> controlled companies cannot do flag-days for large-scale deployment
> changes.

I agree it's not entirely practical, but as there is no easy or fast way
to deploy new RR types, we're limited to the non-practical.  Your point
is well made.  Maybe "flag day" is overstated, but a predefined
recommended phase-out schedule would be a good idea if the TXT record is
optional.

Daniel

-- 
Daniel Quinlan
http://www.pathname.com/~quinlan/