Dear Hameed

 

There is no simple solution to this problem. I think a list of confusables for all scripts is required to be generated then if a domain name is registered using one of the confusable it can not be registered with its counter part:

 

For example if there is already a www.apple.com with cyrillic letter 'a' then one should not be allowed to registered www.apple.com with latin letter 'a'. This will effectively solve the homoglyph problem at GTLD level. 

 

Similarly a list of case fold and ligature normalization confusables can be generated. For example one instance in this list can be Ø¢  (Alef with maddah above U+0622) and ا+Ù“ (Alef + maddah above; U+0627 + U+0653). Both should be allowed at GTLD level but again FCFS (First come first serve) policy should be implemented for instance if www.آج.org is registered with 0622 then it can not be registered with 0627+0653. This makes it a universal solution. These can not be normalized into a single character because 0627+0653=622 in Urdu but not in Sindhi.  This is not a very convincing example but I am unable to think of another one right now.

 

At CCTLD  level language table solution can be implemented or perhapps we can have the same solution discussed above for the sake of orthoganality.

 

Regards

Nadir

 

On 10/3/07, Sarmad Hussain <sarmad.hussain@...> wrote:

Dear Hameed,

 

That is precisely a problem which we need to find a solution for.  One "easy" solution is to have a script centric language(?)-table.  However, if that is allowed, it enables users to type different similar-looking characters (0649 vs. 06cc) which can cause security problems. 

 

However, if language-table is truly LANGUAGE table, then multiple languages will share the character set.  Latter could be the preferred solution.  In this case if multiple languages use the same URL, then it would be first-come-first-served (which is what the model is currently, in any case).  Thus, if you register www.kabulbank.com then I cannot get it afterwards.

 

I am cc:ing others who can also comment/advise on this, e.g.  Ram has experience from Indian langauges which share similar script/langauge challenges. 

 

Regards,
Sarmad

 

 

 

 

---

From: Hameed Sherani [mailto: hameed.sherani@acsa.org.af ]
Sent: Wednesday, October 03, 2007 11:33 AM
To: 'Sarmad Hussain'; 'Seemin Suleri'
Cc: 'Maria Ng Lee Hoon'
Subject: IDNs

 

Dear Dr Sarmad:

 

What about a domain name that appears to be with the same characters in Urdu Persian Arabic and Pashto i.e www.kabulbank.com . Do we have any reservations in this.

 

Thank you.

 

Hameed Sherani| Project Manager| ACSA| CPL (Country Project Leader) PAN Localization-Afghanistan Component| 13th Floor Ministry of IT and Communication Kabul Afghanistan| hameed.sherani@acsa.org.af haidersherani@ yahoo.com|