[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Heirarchical error codes
> From: "Frank Dawson" <firstname.lastname@example.org>
> Date: Mon, 22 Mar 1999 08:36:56 -0500
> On Thursday, March 18, 1999 8:29 PM, Doug Royer wrote in part:
> >> >Doug and Frank
> >> >responded that that wouldn't happen, because all error codes have to be
> >> >registered; there aren't any X- errors. The point, however, is that
> >> >registration doesn't help when you're talking to an old client that
> >> >doesn't know the new error codes.
> >> So, you think that we should support "older" CUAs connecting to "newer"
> >> Isn't this fraught with problems?
> >> At the time you connect, a CUA will get the CAP version number of the CS,
> >> right? So, the CAP CUA will begin by knowing that it is talking to an "up
> >> version" CS. Won't this key the CUA that it may get different
> >> codes?
> >I think this is a flaw with IMAP and we should not do this.
> >The only way that most customers would find this is when all of a
> >sudden - their CUA stopped working?
> By this same logic, an arbitrary CUA should be able to speak LDAP or IMAP.
I don't reach the same conclusions.
> If a CUA connects to a CS and doesn't find the server-side protocol
> environment that it is compatible with, then this is a fatal exception and
> the sessions shouldn't continue.
Yep - its busted as I see it.
In the future a CS server might say only V2. Then V1 <-> v2 is busted.
If however we specify that if a CS wishes to be backward compatible
with V1. There needs to be a way for the CUA to say that. If we
don't define it now, it will have never been in the CUA-v1 protocol
for it to have been said. Then there will be no hope of a V2 CS
being told to operate in V1 mode.