Conflicting applicability requirements in the Bidi Rule

[Peter Occil]

In the case of domain names containing right-to-left characters, the requirements in the Bidi Rule and those of the rest of IDNA2008 seem to conflict.

RFC 5893, which defines the Bidi Rule, says that the rule "applies to labels in Bidi domain names", which are defined as domain names with "at least one RTL label", that is, "a label that contains at least one character of type R, AL, or AN".  Therefore, ASCII-only labels (and particularly those that start with a digit, which do not satisfy the Bidi Rule) that occur in Bidi domain names are subject to the rule. 

(Note that the rule itself redefines "RTL label" and "LTR label".  In a future update, it would be worth rewording the rule so as not to use these two terms.  But this issue is not as important as the applicability conflict.)

However, RFC 5891 sec. 4.2.3.4 says, "If the proposed label contains any characters from scripts that are written from right to left, it MUST meet the Bidi criteria".  (Note that this does not necessarily mean "characters of type R, AL, or AN".)  Therefore, even ASCII-only labels that occur in a domain name with another label with right-to-left characters would not be subject to the Bidi rule.  RFC 5890 sec. 2.3.2.1 also says that a U-label is "subject to ... the Bidi constraints in [RFC 5891] if it contains any character from scripts that are written right to left."

But that conflicts with the statement in the Bidi Rule that "even ... pure ASCII labels have to be tested" for conformance to the rule, and pure ASCII labels are not RTL labels.

And the Bidi Rule conflicts with the statement in the same section of RFC 5890 that says that "IDNA is defined for labels, not for parts of them and not for complete domain names", and the Bidi Rule applies to "domain names", not "labels", at least in the sense that labels in only some domain names are subject to the rule.

(I am aware that some pure-ASCII domain names contain labels that start with a digit.  Those domain names are unaffected by this conflict.)

I would like some discussion on how to resolve this conflict best.

--Peter
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.alvestrand.no/pipermail/idna-update/attachments/20140422/c541cfeb/attachment.html>