On 06/ 1/10 11:26 PM, Pascal Thubert (pthubert) wrote:
Hi Richard:
The trick is that the abstract data structure used for both is really
similar. So it is highly tempting to describe them as a unique
thing.
But the lifecycles of a cache and a registry have nothing in common.
So I proposed to describe them as 2 different abstract entities and
then explain their interaction should they both be present in a same
router, like glean a cache entry from a registration or remove a
cache entry upon optional NUD.
Obviously, this is all abstract and my goal was to make it the
operations clearer to the reader, not to constrain an implementation.
More in the thread "proposed changes for ND".
The conceptual model is that there is only one data structure, Hence
there is no need to describe their interaction.
I suspect having a conceptual model with two data structures and the
associated description of their interactions will most likely lead to
more complex implementations. In any case it is more complex to describe
that model. And it doesn't seem to add any benefits.
Of course, as long as the externally visible behavior is the same as in
the conceptual model, an implementation can use two or more data
structures to represent what is conceptually one data structure.
Erik
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
