On 6/28/21 10:58 AM, Ulrich Windl wrote: Strictly speaking: Nope. See definition of RDATA: https://datatracker.ietf.org/doc/html/rfc1035#section-4.1.3 Anyway use of DNS RRs is specific to application protocols. Thus broad statements like yours are always wrong. Also this general statement is wrong. This is only applicable when resolving host names or MX RRs. Are you actually overseeing what you're talking about? What's "needed for LDAP" depends on how certain LDAP clients (e.g. MTAs) make use of it. And that's why some people are writing specifications. Ciao, Michael.

Thanks Michael. As it happens, I had also come across your draft submission for an EAI-compatible mail attribute along with those I mentioned originally. :D At work I've been looking into EAI support for an e-mail solution which relies on OpenLDAP as its directory software. I've been quite dismayed at the lack of work in that direction on the LDAP front (and what appears to be lack of recent standardization work on LDAP in general). On top of that, from what I've seen, there has been quite some work on UA topics in RFCs across several areas, all more or less suggesting or mandating NFC normalization while LDAP ones, and OpenLDAP from what I can see in the code and empirical testing, go for NFKC normalization, something I think will be rather problematic for us going forward. Since there is effort going into Universal Acceptance (see https://uasg.tech) I imagined that more work would have gone into internationalization for LDAP as well, but reality appears to be different at present. As you said, it's not something just confined to schemas. Btw is there some kind of rationale available as to the direction of certain design choices, for instance for the normalization thing I've mentioned previously, as far as you know? Thanks a lot for sharing. Fulvio