Re: KEYDATA was Re: ipv6 was Re: LC model draft draft-ietf-forces-model-07.txt

Thanks.  I will be looking at using most of your suggested wording 
when I tighten up that section.
I believe all of your clarifications are correct.
I will also try to be more consistent about index vs subscript and 
table vs array.  As you say, it is much clearer if we refer to a 
single thing in a single fashion.

With regard to the specific question of entry assignment,  when a CE 
creates an entry it normally picks an element to assign.  Given that 
no matter how we assign things, there can be deletions, arrays can 
have holes (indices which do not have elements, while the indices 
before and after them do have elements.)
We considered having arrays collapse the holes themselves, but 
decided that would be an actively bad idea.
That is why the array properties include the first used and first 
free index, so that the CE can find things if it loses track for any reason.
I believe that the protocol also has an operation that simply creates 
an entry in an array.  This results in using the first available 
entry, without the CE needing to identify that.  The response 
includes the index of the created entry.  (But that is a protocol 
shortcut, not a model requirement.)

Yours,
Joel

At 12:04 PM 10/30/2006, tom.petch wrote:
>I then wonder who creates the index value, are they
>consecutive integers, is the value for an array entry fixed for the 
>life of the
>array entry?