On Feb 1, 2011, at 4:22 PM, David Hankins wrote: > Thank you, I will wait a bit and see how Ted responds before making any > changes. For the moment we have what is hopefully a compromise between a > two-person consensus.
You specify "MUST" when you want the implementation to enforce the behavior. We do not want implementations to have to support option-specific behavior. Hence, we should not use MUST. However, this is not to say that implementations receiving these options should be required to deal with erroneous results; it's simply that the server should be allowed to return extra data if it's configured to do so; the consumer of the data (not the DHCP client, which is merely a transport mechanism for the data) should be responsible for making sense of it, or rejecting it if the server was configured to send something bogus. _______________________________________________ Softwires mailing list Softwires@ietf.org https://www.ietf.org/mailman/listinfo/softwires
