We must recognize that we are not starting at ground zero. There are
already many Arabic domain names. What your proposal would mean is removing existing ones. The same goes for Devanagari (I think you meant that instead of Sanskrit) which is used for Hindi and a number of other languages. Removing these would affect many hundreds of millions of people, and be, I have no doubt whatsoever, not greeted with acclaim in the developing world. If you want to make the newspapers -- and not in a positive way -- this would
<span style="font-style: italic;">really </span>do it!<br><br>All of the issues you cite are well understood, and ZWJ/ZWNJ/trailing BIDI Mn are narrow extensions of what exists already. Moreover, I think we have to be clear about the impact of these changes. For example, the absence of ZWNJ is actually not an issue for most of the languages using the Arabic script. It is, however, an issue for some, like Iranian. Removing the ability to use the Arabic script for ALL users, because of ZWNJ, is complete overkill. It's hard to draw a precise analogy with Latin, but suppose that IDNA didn't support 'x', and someone said that because of that, we need to retract all IDNs using the Latin script.
<br><br>The current absence of ZWNJ is a handicap for Iranian for expressing certain words, but I can't imagine that any Arabic-script users would be happier if you retracted all ability to use the Arabic script by all users because of this issue. It'd be like closing all of interstate 5, because until you can fix a blocked exit in Sacremento. Better to let people use the freeway,
<span style="font-style: italic;">and </span>work to unblock that exit.<br><br>Mark<br><br><div><span class="gmail_quote">On 12/19/06, <b class="gmail_sendername">Harald Alvestrand</b> <<a href="mailto:harald@alvestrand.no">
harald@alvestrand.no</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><br><br>--On 19. desember 2006 14:25 -0800 Mark Davis <
<a href="mailto:mark.davis@icu-project.org">mark.davis@icu-project.org</a>><br>wrote:<br><br>>> Many, including Arabic, Sanskrit and Dhivehi. Possibly Hebrew too. But<br>>> "leaving out" may be an underspecified term here - see next comment.
<br>><br>> Your statement pretty much floored me. Before we remove the ability to<br>> use domain names from billions of people, it'd be good to have solid,<br>> defensible reasons for doing so.<br>><br>
<br>Let me complete my sentence....<br><br>Until we have a decision and an algorithm that we are sure makes sense for<br>the use of Arabic modified forms, Arabic vowel marks and Arabic shaping<br>modifiers, I think it makes more sense not to register any Arabic domain
<br>names.<br><br>Until we have a decision and an algorithm for determining when ZWJ/ZWNJ<br>should be allowed in Sanskrit domain names, I think it makes more sense not<br>to register any Sanskrit domain names.<br><br>Until we have a rule that allows us to use the vowel marks in Dhivehi
<br>without causing damage to any other part of the registration set, I think<br>it makes more sense not to register any Dhivehi domain names.<br><br>As soon as we know that we have a rational decision on the known issues
<br>with a certain script (ISO 15924 meaning of the word), and a reasonable<br>confidence that there are no more issues about to bite us, I'm all for<br>allowing registries to say "our policy is to allow registrations that use
<br>characters from this script (Unicode meaning of the word)".<br><br>In *all* these cases, I think it makes sense to tell IDNA implementors on<br>the *lookup* side that they should allow those characters to be encoded; if
<br>an user types them, they should go across the wire. The worst that will<br>happen (in the absence of client-side "mappings") is a "no such domain<br>name".<br><br> Harald<br>
<br><br></blockquote></div><br>