On Sat, Jan 28, 2023 at 10:49:45AM -0800, Rob Sayre wrote:
> Viktor Dukhovni wrote:
> > Finally, it is still unclear why any of this is an issue for this work.
>
> I agree that this is a valid, if somewhat uniformative, way to approach the
> issue. That's why I mentioned removing all of it as an option.
>
> But, the draft as it stands still has Section 2, and the later IDNA text.
> https://www.ietf.org/archive/id/draft-ietf-uta-rfc6125bis-10.html#name-identifying-application-ser
>
> You can't follow that advice and get the interoperability people expect
> from the IETF. As you wrote yourself, Postfix just used a ubiquitous
> library and ended up with UTS-46, without really having an opinion on the
> matter. That says it all to me.
Given unclear interoperability for names valid under UTS-46 that are not
valid under IDNA2008, it seems not unreasonable for the current work to
not promise interoperability for such names.
If an application (as many likely will) applies the rules specified here
to a broader set of names, it may interoperate with servers that obtain
certificates with names outside the IDNA2008 space. Other applications
may reject UTS-46\IDNA2008 (set difference) names.
The user can always specify the domain in A-label form from the outset,
in which case many applications will entirely ignore both IDNA2008 and
UTS-46 and just use the A-labels as-is, whether or not they decode to
UTF8 forms that comply with either specification.
--
Viktor.
_______________________________________________
Uta mailing list
Uta@ietf.org
https://www.ietf.org/mailman/listinfo/uta
