Re: [Ntp] Antw: Re: Antw: Re: Antw: [EXT] Éric Vyncke's No Objection on draft-ietf-ntp-mode-6-cmds-09: (with COMMENT)

[Harlan Stenn <[e-mail]>]

On 8/24/2020 4:59 AM, Ulrich Windl wrote:
>>>> Harlan Stenn <[e-mail]> schrieb am 24.08.2020 um 13:41 in
> Nachricht
> <[e-mail]>:
> 
>>
>> On 8/24/2020 3:34 AM, Ulrich Windl wrote:
>>>>>> Harlan Stenn <[e-mail]> schrieb am 24.08.2020 um 10:52 in
>>> Nachricht
>>> <[e-mail]>:
>>>> On 8/23/2020 10:52 PM, Ulrich Windl wrote:
>>>>>>>> Éric Vyncke via Datatracker <[e-mail]> schrieb am 21.08.2020
> um
>>>>> 16:32
>>>>> in Nachricht <[e-mail]>:
>>>>> ...
>>>>>>
>>>>>> -- Appendix A --
>>>>>> Is there a reason why the mode 7 is in the appendix and not in the main
>>> body
>>>>>
>>>>>> ?
>>>>>
>>>>> Because it's a different (and really obsolete) protocol. At least per
>>>>> specification, maybe not by implementation.
>>>>
>>>> No, it is not obsolete.  Mode 7 is for vendor-specific operations.
>>>
>>> You mean mode 7 has been "reassigned"? Otherwise you contradict to what
> you
>>> were postulating many times before.
>>> It seems I don't quite understand: If leaving mode-7 to solely
>>> vendor-implementation, there will be a true protocol mess.
>>> At least some framework to implement vendor-specific operations is needed
>>> IMHO.
>>> So a server would hopefully recognize at least that it does not understand
> 
>> the
>>> command being received and return an error message that the client can
>>> decode...
>>
>> Sorry, Ulrich - I don't know what you're talking about.
> 
> Hi!
> 
> it's about "Mode 7 is for vendor-specific operations.". I read it that
> different implementation may use mode-7 to implement arbitrary protocols with
> that, while currently there's (AFAIK) exactly one "vendor" that had an
> implementation. So not to make matters worse, I'd make this obsolete now,
> allowing to redefince that in a useful way later for NTP version 6 or so.

You mean the reference implementation, and by far the dominant installed
instance?

For many years' time, there were command/control options that were only
available via mode 7.   About 10 years' ago we made sure that most
anything one could only accomplish with mode 7 directives could also be
implemented with mode 6 directives.  While it's still possible to do
these things with mode 7, we're trying to shift them to mode 6.  This
also goes to my point in other messages about how we need to have a
basic set of standardized mode 6 directives, and a basic/minimal set of
data entities that are accessible thru this mechanism.  IMO, there also
needs to be:

- a relatively easy way to augment the minimal element list
- a standard way to learn about the range, format, and description
  of the available elements

I'm not currently in favor of deprecating mode 7 because it might be
useful for something else somewhere down the road.

For example, mode 7 could well be a good way to develop and test
experimental directives that could become mode 6 directives.  There
could well be other uses for mode 7 directives.

> Regards,
> Ulrich

-- 
Harlan Stenn <[e-mail]>
http://networktimefoundation.org - be a member!