Re: [Ntp] Quick review of WGLC for status change for draft-ietf-ntp-update-registries

[Harlan Stenn <[e-mail]>]

On 8/17/2022 2:00 AM, Miroslav Lichvar wrote:
> On Tue, Aug 16, 2022 at 05:27:53PM -0700, Harlan Stenn wrote:
>> On 8/16/2022 6:32 AM, Salz, Rich wrote:
>>>   >    The document discussed here is trying
>>>   >    to update the IANA table to include both sets of values in order to
>>>   >    prevent conflicts with newly specified extension fields, as already
>>>   >    happened with NTS.
>>>
>>> Yes, exactly.
>>
>> Shenanigans.
>>
>> I have asked for specific details about the conflict and nobody has answered
>> me.
> 
> The main problem is that the conflict exists.

Which *still* doesn't answer the root question.

In what scenario does it matter that this conflict exists?

> I agree that in this specific case it can be resolved by sufficiently
> robust NTP implementation. Nobody is saying that it cannot be done.
> But it shouldn't be necessary. If there was a conflict between
> different extension fields, it might not be possible.

Then we have a fundamental difference of opinion.

The cost for avoiding the conflict is high.

You are already advocating a high cost solution to deal with the 
incompatible changes between a v4 and a v5 packet.

If a v5 server only follows the v2 EF format (as described in 5906) and 
does not respond to v4 packets, there is no EF type conflict.

If a v5 server decides to support v4 timing packets as well and does not 
support autokey, there is no EF type conflict.

If a server decides to support newer and older v4 EFs, then based on the 
work I've already done it's clear to me that this "problem" is actually 
No Big Deal.

> The whole point of maintaning the IANA table is to avoid conflicts and
> enable interoperability between different implementations.

Sure, and your definition of "avoid conflicts" is different from mine.

But above, you did say what I described doing was "more robust".

Why are you arguing so hard for the less robust solution?

Your solution places a higher cost burden on future efforts.  For what 
benefit?  The "conflict" you are complaining about is with autokey, 
which does not seem to be in any significant use on the public internet.
-- 
Harlan Stenn <[e-mail]>
http://networktimefoundation.org - be a member!