Comment on Nameprep <draft-ietf-idn-nameprep-10.txt>

Forwarded with permission.

>----- Begin Included Message -----<

Date: Mon, 3 Jun 2002 09:51:10 +0300
From: "Matitiahu Allouche" <[email protected]>
Subject: Comment on Nameprep  <draft-ietf-idn-nameprep-10.txt>
To: [email protected]

Despite some discussion about restrictions that should be set on Bidi
characters in internationalized domain names, none of those have been
integrated in the document.  I therefore propose to add the following
section.

===== Start of proposed additional text =====

5.2 Inappropriate combinations involving Right-to-Left characters

5.2.1 If a domain name label contains any Right-to-Left character (defined
as belonging to Unicode Bidirectional categories R and AL), it MUST NOT
contain any Left-to-Right character (defined as belonging to Unicode Bid
irectional category L).

5.2.2 If a domain name label contains any Right-to-Left character (as
defined above), a Right-to-Left character MUST be the first character of
the label, and a Right-to-Left character MUST be the last character of the
label.

===== End of proposed additional text =====

The objective of these restrictions is to avoid ambiguities (one label
display which may correspond to different character sequences) and
"blending" of text across label boundaries due to the Bidirectional display
algorithm.  In the example below, upper case Latin letters represent
Right-to-Left characters, lower case Latin letters represent Left-to-Right
characters.

Example 1
Storage:   www.travel.madridTOparis.com
Display (assuming a Left-to-Right context):   www.travel.madridOTparis.com
Display (assuming a Right-to-Left context):   paris.comOT.www.travel.madrid

Part of the confusion is avoided by restriction 5.2.1.

Example 2 (assuming a Left-to-Right context):
Display:   www.123ABC.com
Storage:   www.123CBA.com  OR  www.CBA123.com  OR  www.1CBA23.com  etc...

Such ambiguity is avoided by restriction 5.2.2.

Example 3 (assuming a Left-to-Right context):
Storage:   www.abcDEF.GHIjkl.com
Display:   www.abcIHG.FEDjkl.com

The blending of parts of different labels is avoided by restriction 5.2.1.

These restrictions do not prevent all possible confusions and ambiguities
when Right-to-Left characters are used in domain names.  However, this is
the most that can be achieved when the rules are applied to each label
separately, as defined in Nameprep introduction.  They are also very easy
to understand and to implement.

Shalom (Regards),  Mati
           Bidi Architect
           Globalization Center Of Competency - Bidirectional Scripts
           IBM Israel
           Phone: +972 2 5870999  ext. 1202    Fax: +972 2 5870333
Mobile: +972 52 554160

>----- End Included Message -----<