IDNAbis spec

[Alireza Saleh]

Abdulrahman I. ALGhadir wrote:
> Hey,
>
>  
>
> [Quote]: “what if we allow diacritics on the domain name then a domain name like
>
> مايكروسوفت.شركة
>
> Will be different than the
>
> مَايكروسوفت.شركة
>
> Because in the second one there is a diacritic on the first letter.
>
> Although this diacritic is implicit in the first one. 
>
> So this might cause a lot of problems in the domain names registration and owner claims.” [/Quote]
>
>   
I think this is the registry ( Zone owner ) decision to allow or deny 
the usage of certain characters including diacritics, however diacritics 
are part of the some languages. There may be characters ( not necessary 
diacritics ) in a languages that using them may cause problems, in these 
cases the registry can decide to remove those characters from the 
character repertoire for that language.
>  
>
> Well this has been answered in “NSM flow?”
>
>  
>
> [Quote]
>
>   “Moreover, for the displaying order of the labels of a domain name I have tried the following hypothetical domain names:
>
>  
> Husni.حاسب.شركة
> حسني.حاسب.شركة
> husni.حاسب.com
> حسني.computer.شركة
> حسني.حاسب.com
> husni.computer.شركة
> husni.computer.com
>  
> The following is an image of the network order from right to left  for Arabic of the above: 
>  
>  
> It is clear that when we use two consecutive RTL labels separated by dots and followed by one LTR label the display order does not look as it 
> should. The same is true that when we use two consecutive LTR labels separated by dots and followed by one RTL. The question is should we allow such confusion?”[/Quote]
>  
> from draft-ietf-idnabis-bidi-06
>
> [Quote] 
>
> “   o  The sequence of labels should be consistent with network order.
>
>       This proved impossible - a domain name consisting of the labels
>
>       (in network order) L1.R1.R2.L2 will be displayed as L1.R2.R1.L2 in
>
>       an LTR context.  (In an RTL context, it will be displayed as
>
>       L2.R2.R1.L1).”
>
> [/Quote]
>
>  
>
> Well this problem was expected to happen, IDNA uses a UAX#9 Bidi algorithm version-like where some rules have been removed.
>   
Which of the above examples represent L1.R1.R2.L2 ? These cases require 
inter-label checking and the working group came to the consensus that do 
not perform such tests.

-- Alireza