John C Klensin wrote: > Better text is welcome if we can agree on the principle. It > may also be that, if we are going to permit addresses, some > words in the Checklist about preferences for IPv4, IPv6, or > parallel examples would be in order. The principle should be "stay away from IP literals when you can in practice". The hypothetical specification could quote the scream in RFC 822: | Note: THE USE OF DOMAIN-LITERALS IS STRONGLY DISCOURAGED. After that's clear some examples with domain literals are no problem, they are actually desperately needed, because the syntax differs depending on the context: Sometimes square brackets are required, sometimes they are not allowed, sometimes this depends on IPv4 vs. IPv6. When square brackets are required they are typically used as is, but in URIs + IRIs outside of <host> and <ihost> they have to be percent-encoded. A mailto: specification without example for domain literals would be irresponsible, no matter what an ID-checklist says. IMO the SHOULD about using FQDNs instead of IPs in examples is nonsense. The ID-Checklist is the wrong place to tell authors that FQDNs are better than IPs; they are supposed to know this. >>> I think that is supposed to be what we all want around >>> here, isn't it? Yes, your new MUSTard is fine. The old SHOULD about FQDNs vs. domain literals in examples is utter dubious. Frank _______________________________________________ Ietf@xxxxxxxx https://www.ietf.org/mailman/listinfo/ietf