Re: [apps-discuss] CONTEXTJ in TLD DNS-Labels (draft-liman-tld-names-05)

[Paul Hoffman <[e-mail]>]

On Jul 19, 2011, at 11:34 PM, Patrik Fältström wrote:

> On 19 jul 2011, at 21.49, Paul Hoffman wrote:
> 
>> We have already seen the perceived need for these characters in the root zone, and we have not seen any statement of how they can cause harm *in the root zone*. "Phishing" in the root zone, given the horrendous weight of the process for getting new names put in the root zone, is not a threat. Which others do you believe that need to be weighed against the value of the characters?
> 
> Yes, phishing in the root zone. People putting URLs on web pages that you click on.

This is an argument about disallowing labels with non-PVALID at any level, not just in the root. Prohibiting non-PVALID in the root will do nothing to prevent those attacks; an attacker just puts the phishing label in a different part of the domain name.

> It is tons of code easier in various applications to "know" that a code point is either allowed or not allowed in the TLD than having context dependent rules that otherwise is the option.

Such a rule just begs for various applications to "know" that non-PVALID codepoints are illegal anywhere. This was discussed in the INDAbis WG, and the WG decided that the root zone was not special.

> So the question is whether security software can filter out URLs with ZWNJ in the TLD as dangerous or not.


This truly sounds like a re-assessment of non-PVALID in IDNA2008. I'm willing to have that discussion if you are, but trying to say that it somehow applies only to the root seems misguided.

--Paul Hoffman