Re: Maintenance WG: ent of Selections

[tom petch <[e-mail]>]

On 28/01/2021 19:05, Joel Halpern wrote:
> With regard to maintenance, the two areas I am most active in each have
> enough maintenance that having one WG for all the maintenance (in each
> area) simply would not work.  IDR, LSR, and MPLS for Routing.  6man
> (which is even named as a maintenance group) and DHC (and probably even
> some of the others) in Internet.

More generally, the work of an organisation can be divided into project 
and process, and the IETF is no exception.  To the process category, I 
might add TCPM, netconf, netmod.  If a prococol is successful, then, 
almost by definition, you need a process WG for it.

You could wish for a WG Chair to cut their teeth on a project WG before 
moving on to the greater demands of a process one.

Tom Petch

> Yes, the IETF used to have a policy against long lived working groups.
> We found it didn't work.
>
> Yours,
> Joel
>
> On 1/28/2021 11:22 AM, Phillip Hallam-Baker wrote:
>> On Wed, Jan 27, 2021 at 7:05 PM Fred Baker <[e-mail]
>> <mailto:[e-mail]>> wrote:
>>
>>     So voila! IPv4 has existed more than ten years, so we don’t need it
>>     any more... Also DNS, DHCP, TCP, Ethernet, SMTP - wow, that rule
>>     really clears the deck. And heck - even IETF chairs produce internet
>>     drafts.
>>
>>
>> Quite. This leads me to re-suggest a proposal I have made from time to
>> time which is that every area have a DISPATCH working group and every
>> area also have a maintenance group.
>>
>> This has (sorta) happened in security with LAMPS. Faced with the need
>> to update cipher suites across the board, a new WG was formed which
>> has been tweaking every protocol in active use not in active development.
>>
>> The objection generally raised is of course that this allows people to
>> 'mess' with existing protocols adding features that shouldn't be
>> there. But that objection presupposes that the job of the IETF is to
>> control permission for permissionless innovation.
>>
>> One of the reasons some WGs linger is that their function requires
>> continuous tweakage, GSSAPI/KITTEN being an example. Better to have
>> one WG with the purpose of lingering than four lingering past their
>> sell-by.
>>
>>
>
> .
>