Normative language in media type registration

[ioggstream]

Question

Is normative language allowed in media type registration?

[richsalz]

Yes. See for example https://datatracker.ietf.org/doc/html/rfc6838#section-1.2

[ioggstream]

Thanks @richsalz I thought that since in IANA there's no mention of BCP14, and that other organizations like w3c can register media types, this was not a good practice.

I think that in very specific cases, normative language should go into the specification though and need not to be duplicated into the media type registration, otherwise as specs are updated there's a risk for inconsistencies. Does it seem reasonable to you?

[ioggstream]

cc: @jdesrosiers normative can be ok

[richsalz]

I think that in very specific cases, normative language should go into the specification though and need not to be duplicated into the media type registration, otherwise as specs are updated there's a risk for inconsistencies.

Yes, that's reasonable.

[darrelmiller]

Is there any follow-up action to this issue or can it be closed?

[ioggstream]

@msporny are there any guidelines on that? Once you answer, we can close :)

[msporny]

@msporny are there any guidelines on that? Once you answer, we can close :)

There is certainly no normative language that goes into the IANA Media Types registry (I don't think anyone was questioning this, so I'm just stating the obvious). It's also true that RFC 6838 contains normative language. It is also true that specifications from W3C contain normative language in their IANA registration section (Example in JSON-LD 1.1).

I don't know if there is any guidance on authoring normative language for media types and/or their registration (this isn't my area of expertise). That said, I expect that it's common (but don't have the data to prove it) that media type registrations, typically expressed in the specification itself, both at W3C and IETF, do expect to state normative things in their IANA Considerations sections. I don't expect that people pay too much attention to the actual registration emails that the domain experts evaluate.

I also agree with your statements above @ioggstream.

I have no idea if I've actually answered your question, though... did I? :)

[richsalz]

In my experience (I'm one of the TLS designated experts), IANA read the draft and asks us to see if it's sufficient. Email doesn't matter, except perhaps as a heads-up to get the designated experts to review. Perhaps it makes sense to ask Francesco, our AD, for her views.