[DNSOP] New I-D: Transparent Handling of IDNs in Name Resolu tion APIs (draft-rodenhaeuser-idna-transparent-resolution-00)

Dear DNSOP,

I have submitted a new individual Internet-Draft that I believe falls
within this working group's scope, and I would welcome the group's
review -- and any interest in adoption.

  draft-rodenhaeuser-idna-transparent-resolution-00
  https://datatracker.ietf.org/doc/draft-rodenhaeuser-idna-transparent-resolution/

The draft addresses the inconsistent handling of internationalized
domain names at the name-resolution boundary. Because applications are
expected to convert U-labels to A-labels before calling resolution
APIs, that conversion is today performed by many different components
following different IDNA standards, using different libraries, and
shipping different versions of the Unicode mapping tables. As a result,
the same input can resolve to different registrable domains depending
on which component converts it. For example, "STRAẞE.de" is converted
to "strasse.de" by a stack whose tables predate Unicode 15.1 and to
"xn--strae-oqa.de" by a current one -- two independently registrable
names, on one host.

The draft specifies that name resolution services accept Unicode by
default and convert exactly once, at the point where a name is handed
to a specific protocol. The wire format is unchanged (A-labels for the
public DNS); the procedure is UTS #46 nontransitional with a Unicode
15.1 table floor. This is not new architecture: RFC 6055 recommended
precisely this placement of the conversion in 2011, and Windows,
systemd-resolved, and the browser platforms already convert by default.
The draft therefore targets BCP, to codify deployed practice.

I would be glad to present this at a future DNSOP session (I can attend
remotely) if the chairs think it useful. Feedback on the list is very
welcome.


Thank you,
Luca Rodenhäuser

_______________________________________________
DNSOP mailing list -- [email protected]
To unsubscribe send an email to [email protected]