Charles Lindsey wrote:
Yes, I know all that. You have demonstrated that a problem could,
_in_principle_ arise. But my question, which I repeat again and which you
still have not answered, was:
Have you heard of any case where this esoteric change of semantics
actually caused a problem?
That problem does occur for anybody attempting to write any
software which involves comparisons of such identifiers, which
is exactly the situation initially posed by Jacob. A choice
must be made between comparing identifier right-hand sides
as domain names (viz. case-insensitive) or as RFC 2822 id-right
(2822 doesn't say whether that's case-sensitive or not). Under
the circumstances, comparing as case-insensitive domain names
seems the most reasonable course of action; it is consistent
with the 20+ year history of identifier RHS as domain names and
it is consistent with RFC 2822's recommendation to use domain
names.