[apps-discuss] IANA considerations in a tangle

[t.petch <[e-mail]>]

This list has provided a home for a number of considerations of 'IANA
Considerations' so I hope it may provide inspiration here.

RFC4379 set up a registry
http://www.iana.org/assignments/mpls-lsp-ping-parameters/mpls-lsp-ping-p
arameters.xml
of TLVs (functions) and sub-TLVs (parameters)
with each TLV having its own namespace of sub-TLVs.  The registry has
expanded in both directions, with new TLVs being added and new sub-TLVs
being added to existing TLVs.  Last year,
draft-ietf-mpls-ipv6-pw-lsp-ping added, and changed, the sub-TLVs of TLV
Type 1, while draft-ietf-mpls-return-path-specified-lsp-ping added a new
TLV, Type 21, and for it, duplicated the existing sub-TLVs of
TLV Type 1, an early allocation of which is still visible in the IANA
Registry.  A later revision of this I-D switched to referencing the then
existing sub-TLVs of the Type 1 TLV.   This is an active area so more
changes are likely -
e.g. draft-zjns-mpls-lsp-ping-relay-reply may do so.

The question that seems not to have been noticed (until pointed out by a
lurker on the list), despite this being an active area, was whether or
not the ipv6-pw changes to the sub-TLVs of Type 1 should be carried
across
to Type 21 or not, and equally, whether reusing the Type 1 sub-TLVs for
Type 21 only applied to those then in existence, one of which ipv6-pw
was deprecating, or should also be applied to any future updates to the
sub-TLVs of Type 1.

The ipv6-pw I-D is currently in with the IESG with a DISCUSS suggesting
that the relevant part of the subject material belongs in a different
WG, which would make the chances of spotting such interactions yet more
remote:-(

With the benefit of hindsight, RFC4379 might have split the range of
sub-TLVs into a part common to all TLVs, and a part unique to a
particular TLV; but it didn't and it would be impractical to retrofit
that now.

How could the IANA considerations be modified to alert those amending
the
sub-TLVs of Type 1 TLV to the fact that they are used elsewhere, and to
alert those reusing the sub-TLVs of another Type of TLV to the fact that
a registry of sub-TLVs is volatile and could be added to or deprecated?
Since this (almost) did not happen a single list while this topic is
still
distinctly active, I expect that it will recur.  Or do I worry too much?

Tom Petch