On Thu, 28 Dec 2006, Brian Haberman wrote:
Given that 2461 (and 2461bis) are DS, I would find it very disturbing if
implementers did not treat the entire document as normative. In order
to be compliant with a spec (any spec), an implementation MUST adhere to
all aspects including protocol constants. Otherwise, how would we ever
have interoperability? I do not see any benefit in having any
specification state *which* components of the document are normative.
Indeed -- but maybe I was not explicit enough. I see three things
here:
1) With most of these constants, there is no interoperability issue
if an implementation were to choose different values. Quickly
looking, I wasn't able to see any defined constant that the
current protocol would require to be exactly the way it is for
interoperability.
Instead, this is more of an issue of harmonizing protocol
behaviour across implementations, not so much about
interoperability.
2) Strictly speaking, if constants are normative, if an
implementation has (say) REACHABLE_TIME of 10 seconds (three times
"faster" than the spec) it should be considered incompliant.
3) in some other protocols what we have defined as constants could
very well have been defined as suggested default values (for
example).
This is more than strictly required for interoperability.
--
Pekka Savola "You each name yourselves king, yet the
Netcore Oy kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
