[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: FW: [Entmib] Minutes from SF (fwd)



--On tirsdag, april 22, 2003 16:07:02 +0200 "Wijnen, Bert (Bert)" <bwijnen@lucent.com> wrote:

Question....
  Requiring implementation of mandatory objects is fine -- that's
equivalent   to requiring the implementation of all MUSTs in non-MIB
specifications,   but requiring implementation of conditional mandatory
is like requiring   implementation of all SHOULDs in non-MIB
specifications.  Does the   IESG really require that SHOULDs be removed
if they have never been   implemented??  For example, if a protocol
specification specifies the   action an implementation should take in
some unlikely scenario, must   that action be removed from the
specification if no implementation ever   encountered that scenario ??
Surely, we don't want specifications with   loopholes even if no
implementation has yet encountered those loopholes.

This came up on context of Entity MIB WG. Full text is below.

My udnerstanding is that the answer is YES. We MUST see an implementation
report where at least two genetically independent implementations DO
implement all the SHOULDs. We do not require that everyone implements
them, but at least two must do so in oder to advance from PS to nay
higher level.

Am I correct or do people have differing opinions?
That is my interpretation too.

The reasoning is that unless we have two interoperable implementations of a feature, we can't tell if it was specified well enough for interoperation - and that question is actually independent of whether or not it's needed in all implementations (which is the difference between MUST and SHOULD).

By the same logic, all the MAYs must be implemented or removed too.

Harald