- From: Ned Freed <Ned.Freed@innosoft.com>
- Date: Wed, 24 Mar 1999 18:49:38 -0800 (PST)
- To: Harald Tveit Alvestrand <Harald@Alvestrand.no>
- Cc: Scott Lawrence <lawrence@agranat.com>, IETF Applications Area Discussion List <discuss@apps.ietf.org>
> The MDN spec has criticality flags. > So does X.509, which is referenced by the PKIX specs. > So I guess we do have precedent for them. Also X.400, which offers a pretty good precedent for how bad such flags can be, even when they are designed into the core protocol elements in a consistent way. I have many horror stories... Conformance to the MDN specification in this general regard is also proving to be a problem operationally, so much so that it has been seriously suggested that it be removed from the standards track. (Note that I neither think this is going to happen nor want it to; I'm simply reporting what has happened.) I don't have much experience with X.509, but my guess is that we're going to see all sorts of interoperability problems surrounding the criticality flags in it, especially as the increasingly obscure options are exercised. Criticality flags are a nice idea that for some reason just don't connect with typical human behavior in the right ways. As such, they are a facility you're well advised not to use in a protocol unless you have an overwhelming need. Ned
Received on Wednesday, 24 March 1999 21:58:54 UTC