On Mon, 2003-12-15 at 19:17, Cyrus Daboo wrote:
[good catch snipped]
I see several possibilities to this:
1) vacation address comparisons are implementation defined: implementations
know whether they do case-sensitive or case-insensitive address matches on
delivery so they pick the appropriate comparator. Some statement of that
should be in the spec.
but the case sensitive system may have expanded the alias and forwarded
it to a different host which isn't.
2) vacation addresses are explicitly defined either case-sensitive or
case-insensitive on the local part - we have to pick which and state in the
spec.
it would have to be case-sensitive, wouldn't it? it's possible to
simulate case-insensitivity, however awkward it is, the reverse is not
true.
3) A new comparator argument is added to the addresses parameter to allow a
comparator to be explicitly chosen - we would have to pick the default in
the absence of that argument.
I think this is the Correct(tm) solution.
--
Kjetil T.