At 11:52 PM 5/5/2002 -0700, ned(_dot_)freed(_at_)mrochek(_dot_)com wrote:
>Agreed. Additionally, the optional fields in 3192 and 3193 are associated with
>offramp functionality, which makes the concerns about being able to map the
>local-parts of such addresses to mailboxes moot.
the purpose of 3192/3193 was for offramp, yes. however that's really a
'local' decision of the system processing the local-part.
what makes the template inappropriate for other sorts of services, some of
which are not offramp, but instead something on the targeting host?
It is approprate as long as interpretation is left up to the system that
"owns" the local-part.
What's can be problematic is the combination of a parameter structure and
encoding of non-ASCII characters. This sort of encoding implies some level of
interpretation of the local-part going on in the context of systems that do not
"own" it but want to "display" it.
There's a slippery slope lurking in this neighborhood. I don't think we're on
it yet, but we do need to be careful.
Ned