[Idna-arabicscript] Re: ZWNJ and CONTEXT

Alireza Saleh saleh at nic.ir
Tue Apr 8 21:56:05 CEST 2008

About the bundling I agree with Behdad's comments which we had a very 
long conversation during the last meeting of  ASIWG in Dubai, but this 
is the higher level solution which is in the scope of the registry as I 
mentioned. This is the exact solution that all other folks at the 
meeting wanted to do with other confusions such as YEH and KAF. Besides 
we need to think about more general solution which I think we can make a 
rule base on characters as well as their categories at protocol level 
which may limit the usage of ZWNJ in third and more level domains.


Behdad Esfahbod wrote:
> On Mon, 2008-04-07 at 22:42 +0500, Sarmad Hussain wrote:
>> ZWNJ is required and should follow one simple rule:
>> The character PRECEDING ZWNJ should be one of the joining characters,
>> otherwise it should not be allowed.
>> I will provide a list of these characters possibly tomorrow.  Note
>> that the character FOLLOWING ZWNJ is not important.      

> This is wrong.  The sequence <BEH, ZWNJ, DIGIT ONE> renders the same as
> <BEH, DIGIT ONE> and the character preceding ZWNJ is a joining
> character.
> I think the best way to handle ZWNJ is to bundle it with the nul-string.
> That is, if someone wants to register a domain with ZWNJ in it, give it
> to them, but also bundle/block/reserve/suggest the domain with all ZWNJs
> removed.  This fixes all the phishing issues with it, gives users the
> domain they want, is what we are going to do with all the other
> confusing groups of characters, and is simple.

More information about the Idna-update mailing list