Re: [alto] High level comments on draft-roome-alto-unified-props-00

[Wendy Roome <[e-mail]>]

Richard,

Thanks for the comments.

1. I do not see how we could define a universal hierarchy; I think the
entity hierarchy and inheritance must be defined in the context of each
specific entity domain. Section 2.3 allows for hierarchies, but requires
the domain to define the rules:

   Addresses may be hierarchical, and properties may be inherited based
   on that hierarchy. Again, the rules defining any hierarchy or
   inheritance must be defined when the domain is registered.

If you can suggest a general way to define hierarchy & inheritance for
arbitrary domains, please let me know.

2. I would say the default is no consistency across domains. If there is
any inter-relationship or inheritance between entities from different
domains, the domain rules must define it.

Because PIDs are defined by CIDRs, some readers might think the two
domains are related. So in 3.2.4 I explicitly said they are not.

	- Wendy Roome

>
>Date: Tue, 21 Jul 2015 16:29:09 +0800
>From: "Y. Richard Yang" <[e-mail]>
>
>Wendy, all,
>
>I just read draft draft-roome-alto-unified-props-00 and liked it a lot. It
>is very well written, as Wendy always does, and I recommend that many of
>you read this design.
>
>Personally, I see this design as a good candidate as a WG item. Before
>making up my mind, I see benefits in discussing two high level design
>decisions:
>1. Hierarchy of general domains. In the current design, this issue already
>appears in the ipv4 and ipv6 domains. The approach that the draft adopts
>is
>longest prefix matching (LPM); see Section 3.1.3.. This can be considered
>as smallest containing set, if we see each CIDR as a set, and such sets
>form a directed acyclic graph. Q: Does it make sense for the document to
>go
>as far as defining this general principle, instead of the specific LPM?
>
>2. Consistency of the same property across domains. Section 3.2.4 gives
>one
>example of such a case. Q: Is this a specific decision for two specific
>domains, or the general principle is no across domain consistency?
>
>Cheers,
>Richard