Re: my response to the proposed changes

I believe Stuart is arguing about names and whether they suggest the =
right thing in the minds of the reader - not about algorithms.

The name "DEFEND_INTERVAL" is used in the current draft in section =
2.2.1. (There is a proposal to change this name to ANNOUNCE_WAIT). For =
this reason re-using "DEFEND_INTERVAL" for what is currently WATCH_WAIT =
is a bad idea however apt the name might be since it is obviously =
leading to huge confusion.

Stuart says of WATCH_WAIT:

> The ONLY relevance of the ten-second time constant is that it imposes =
a=20
> rate-limit on the frequency of defensive ARPs that are allowed, to=20
> prevent storms.

Not the only relevance - this period defines a frequency of conflict =
above which a host is obliged to stop defending and find another =
address. This goes beyond storm prevention and defines a practical test =
for whether an existing address is tenable.

> Sensible names might be:
>=20
>   DEFEND_INTERVAL
>   MINIMUM_DEFEND_INTERVAL
>   DEFENSIVE_ARP_RATE_LIMIT
>   RETALIATORY_ARP_RATE_LIMIT
>   MANDATORY_RECONFIGURE_WINDOW

Try "DEFENSE_WINDOW" - it's snappier.

I support Stuart's assertion that WATCH_WAIT is not a good term. For the =
reasons above, I do not support the use of DEFDEND_INTERVAL for =
anything.

Philip