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

Re: I-D Action:draft-ietf-sasl-channel-bindings-00.txt





On Apr 14, 2009, at 4:24 PM, Kurt Zeilenga wrote:

It is the nature of SASL that such introductions will likely have an impact upon implementation-specific abstractions to hide the details of mechanisms from protocol implementations.

For instance, RFC 4422 is mum on the need for a framework implementations to manage information about identities established a lower level in order to support EXTERNAL. And with the introduction of EXTERNAL-*, the framework implementations will likely need to change. Instead of managing a single external identity (as most seemed to have done), they will need to manage one per each of the EXTERNAL-* variants. Today, it seems that the community thinks this will be enough. But our crystal ball in not 100% clear. Maybe in 5 years, someone will want to introduce a mechanism that requires framework implementations to provide something different.

As RFC 4422 doesn't itself attempt to abstract away the differences between mechanisms, it can and should remain mum here.

Likewise, RFC 4422 ought not try to abstract away the differences in mechanisms that support channel bindings. For instance, presently the suggestion is mandate two names per mechanisms. But what if tomorrow we find this isn't good enough. We'd have to update RFC 4422 to allow introduction of something which was just because we were short-sighted today.

I do think it appropriate to encourage future mechanisms to be similar to existing mechanisms where sensible. What I object to is us mandating future mechanisms be the same in area based upon what we think is sensible today.

RFC 4422, I think, tries very hard to leave the future open. This proposal, I think, tries very hard to close our future.

-- Kurt