In your previous mail you wrote:
> In your previous mail you wrote:
>
> I don't know if there is a requirement for the stability of ids
> that has bearing on this suggestion.
>
> => I believe there should be none because the stability requirement
> should be for id names, not id values. Of course, we should recommend
> to use always names too.
Notice that I did *not* talk about stability across reboots.
My issue is whether it is ok or not to have the ids change
on a whim while the machine is *running* (e.g. due to using hotplugging to
add or remove NICs).
=> it seems we agree to say it is *not* ok.
If we allow this rate of change it seems like an application
needs to do the name to id translation each time it is using
an id for instance each time it is sending a packet.
Is that really what we want?
=> no, we don't want that. Only the inheritance stuff should be called
each time. Note we don't want to change the name-id file very often too.
Regards
[EMAIL PROTECTED]
PS: I believe you'd like to get some freedom in the way zone ID values
are chosen. If we use a per interface list of zones which the interface
belongs to, then we can implement what we'd like... Simple schemes are
easier because they don't need the reverse map (zones to interface lists).
PPS: simple schemes can be described and simulated when the name-id file
is configured, do more complex schemes need a standard way to express
zone/interface mapping? I believe the only hairy case is the mobile node one,
so do you know what we need for such cases?
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
