Re: Heirarchical error codes

["Frank Dawson" <fdawson@xxxxxxxxxxxxx>]

On Friday, March 19, 1999 9:53 AM, Rob Earhart wrote, in part:

>  I think it's better to avoid limiting things to a linear series of
>protocol versions; there will be different extensions, and they will be
>orthogonal to each other.  It seems more flexible for each side to
>announce a list of extensionss they support, and then both sides can
>behave appropriately.


Rob, how do you see this working? It seems pure chaos if both the CUA and
CAP can do whatever they want in terms of behavior. This CAP protocol will
only work if we nail down the semantics of what a CUA must do and what the
CS must do with regard to "extensions".

We are talking about "standardized" extensions here, right? The discussion
so far did not leave any room for non-standard, experimental extensions.
These beasts smell like real interoperability problems-in-the-making to me.

Can you explain more of what you envisioned here?

- - Frank