Alexey Melnikov writes:
Dave Cridland wrote:A simpler solution might be to promote SASL PLAIN as the i18n plaintext login, since that either will be, or has been, fixed, and deprecate LOGIN for even that.I would rather the document just state that both LOGIN usernames and passwords are UTF-8 now.
Much as I'm in favour of that, I feel that this document is the wrong place for it. This document specifies two extensions. Making an unconditional change to 3501 in the same document feels bad. A bit like the way RFC 2231 modifies RFC 2060: Two lines changing 2060 buried in the middle of a ten-page document.
I suppose section 5 could be beefed up, so it's not such a tiny change. It could allow UTF-8 logins, passwords and mailbox names?
Related to this. The second paragraph says: When AUTHENTICATE is used, some servers may support userids and passwords in Unicode [13]. However, such userids cannot be used as email addresses, and at present also seem to be incompatible with the current latest ACL drafts. Unless the ACL drafts resolve this, server authors are cautioned against supporting ACL and unicode userids simultaneously.The updated ACL got published as RFC 4314. So the end of this paragraph is wrong.
Simplify my life. Give me text ;) Arnt