On 7 mar 2009, at 21.04, David Conrad wrote:
On Mar 7, 2009, at 8:40 AM, Patrik Fältström wrote:The problem with writing exact objective rules is that with the 6000 languages, and enormous number of codepoints, it is extremely hard to create say a regular expression that we know is _absolutely_ correct regarding separating the good TLDs from the bad ones.Agreed.Without knowing the policy for the 2nd level domain, I think it is very hard to say whether a given TLD level is safe or not.Unfortunately, as you're aware, policy at the second level varies over time and there are few (if any) limitations placed upon the evolution of that policy. Attempts by ICANN to constrain 2nd level policy would unlikely be a bit ... challenging (particularly in the ccTLD space).
True...so, and to clarify my earlier statement. I think personally it is absolutely safe to allow TLDs that consists of codepoints that have a so called strong directionality. If the directionality is the same. This will definitely allow both RTL and LTR labels.
Now we can from this add more things, for example, talk about whether other codepoints (like digits) can be part of the label, etc.
But we should be careful, very careful. It is hard to undo such a decision as we all know. And remember that a domain name can exist in various contexts, including of course the ones where there is text "after" the TLD (if you look in logical order), and we might have any text there, like a digit, different directionality etc.
Patrik
PGP.sig
Description: This is a digitally signed message part
_______________________________________________ DNSOP mailing list [email protected] https://www.ietf.org/mailman/listinfo/dnsop
