To:
Edward Lewis <lewis@tislabs.com>
Cc:
ietf-provreg@cafax.se
From:
Dave Crocker <dcrocker@brandenburg.com>
Date:
Wed, 29 Aug 2001 15:59:21 -0700
In-Reply-To:
<v03130316b7b2c7479e1e@[199.171.39.33]>
Sender:
owner-ietf-provreg@cafax.se
Subject:
Re: Pushing vs. Pulling
At 09:23 AM 8/29/2001, Edward Lewis wrote: >BTW, the Push-Pull straw poll did have one benefit. Scott suggested that >polling is a MUST, pushing a MAY. Perhaps this is the middle ground to >build upon. (Comments on this, please.) yes! d/ ps. Occasionally the meaning of must vs. may is confused (and confusing) in IETF discussions. My own understanding of the usage is that MUST means that any conforming implementation must support the capability. However even a must does not mean that the capability must be USED during operation. A MAY means that a conforming implementation is permitted to include the capability, but is not required to. However the mechanism defined by the MAY is the "standards" way of doing the task that is defined. In this case, that would mean that if push is supported by an implementation a standards conformant implementation will perform push in the way defined by the MAY. Just for completeness, it is worth noting that a third type of requirement level in IETF work is in between may and must: SHOULD. SHOULD means that it is strongly recommended that the feature be supported in the implementation -- again note that this does not require its use during actual operation -- but that an implementation may choose not to support it... if there is a very good reason. (That sounds fuzzy, and it is. Essentially, it means that there is strong pressure to support the feature, but you are still allowed to claim conformance without it. On the other hand, you should expect customers to ask you to defend the decision.) Hence, MAY is actually a very modest standards level of requirement. d/ ---------- Dave Crocker <mailto:dcrocker@brandenburg.com> Brandenburg InternetWorking <http://www.brandenburg.com> tel +1.408.246.8253; fax +1.408.273.6464