On Mon, Nov 13, 2017 at 05:47:20AM +0000, Mark Svancarek via UA-discuss wrote:
The assumption is that all local parts are ASCII letters-digits and that IDNs in the domain part should be expressed in punycode. This is of course doubly broken, because humans can’t really use punycode.
Fixing the local part is a good start and I encourage it. But if the domain name part continues to prohibit Unicode, it won’t actually help anyone.
The domain part does not actually prohibit Unicode in the strict sense. There's a note in the document that says This syntax allows e-mail addresses with Internationalised Domain Names using punycode, such as example@xn--d1acpjx3f.xn--p1ai. A user agent should represent that in the user interface as example@яндекс.рф So, the user-agent is asked to do the IDNA transformation from A-label to U-label for display purposes. Since under IDNA2008 that ought to be a 1:1 and fully reversible operation, it shouldn't be a big deal. It's true that this input restriction won't produce an EAI address, but that is trivial to fix, also, if you do the IDNA transformation. Best regards, A -- Andrew Sullivan ajs@anvilwalrusden.com