On Tue, 18 Mar 2015, John Levine wrote:
I'd suggest taking out "based on special characters", since the most common mapping is case folding. The standard term for the LHS is local-part, so you might as well use that and reference RFC 5321, sec 2.3.11 where it's defined. Also, the SHOULD NOT would better be MUST NOT, to be consistent with RFC 5321 which says "the local-part MUST be interpreted and assigned semantics only by the host specified in the domain part of the address." (If you have a private agreement with someone and you have knowledge of their internal mappings, you can do whatever you want, but of course private agreements are outside the scope of standards.)
With this, we've gone a full circle. It feels strange to me to write a MUST NOT, knowing that implementors will need to do this in practise. But I guess I could live with it if the consensus moves this way, but to me that seems only because I know the MUST NOT will be violated.
Having said all that, how useful do people think this will be if it doesn't allow the local-part fuzz that mail systems provide? If the addresses to be looked up are picked mechanically from incoming mail headers, it'd likely work fine. If typed in from business cards, it could be pretty frustrating.
Or even typed in on an iphone for which you have an existing contact with a lower case name, and it stupidly Uppercases it anyway when you type. Paul _______________________________________________ dane mailing list dane@ietf.org https://www.ietf.org/mailman/listinfo/dane