[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Generic TC Module



C. M. Heard wrote:

IMO, TCs should be grouped such that strong cohesion
with other definitions is maintained.

In reality,  it takes way too long to do the right thing
in the IETF, and so WG chairs and authors (like me)
resist attempts to pull TCs and other generic constructs
into a "generic MIB", because it just becomes a normative
reference that will delay the original document.

Case in point -- the TPM and SSPM MIBs in RMONMIB
waited an extra year for the IPPM-REGISTRY-MIB, which
is nothing more than a list of OIDs.


Andy


On Thu, 1 Sep 2005, Wijnen, Bert (Bert) wrote:


I would think that if we declare that the TC first needs to be documented (not in a MIB module, but just as a simple TC) and
that we need IETF consensus on it, then we can do it. It would
be OK (as far as I am concerned) to just have such a document
as Informational. The IESG does a 2 week Last Call on them,
and only after approval will IANA add the TC.


For many simple TCs (like the Unsigned16TC and Unsigned8TC) I
would think that this can be done fairly easy and quickly.
For more complicated TCs I see problems. But if they are complicated, then I doubt they are always genereic. In such
cases we can require them to just be done in some other
document first.


And I think we can defend the idea that the IANA maintained
TCs are good enough for documents at PS, DS and STD level.
We can write that up in our initial document where we set the rules for additions and an initial set of TCs.
Maybe that initial doc should be BCP and it is OK to
have a ref to a BCP from a DS or STD doc. And we can
make it explicit that IETF has consensus and IESG approves
that such an IANA MIB module can be referenced from PS,
DS, STD.


See... I'd like to be practical and just DO it.

But if two of our core MIB doctors (Mike and Juergen)
already have an issue with it... then we won't get far.



If there is consensus to proceed with this idea I certainly won't stand in the way. I don't object to having a "living list" of useful TCs, or even (in principle) to doing all of our standards in this way. I just questioned whether _having the IANA_ maintain these things is appropriate. It somehow does not seem like maintenance of registries of assigned names and numbers, which is the IANA mission.

However, given the lack of action in newtrk in the last couple
of years, I can see the motivation for using a mechanism that
we have in place instead of waiting for years for something
to come out of newtrk, with the significant risk of coming up
empty-handed.

//cmh