Ok, I understand now what you mean, thanks.<br><br>Mark<br><br><div><span class="gmail_quote">On 12/22/06, <b class="gmail_sendername">John C Klensin</b> <<a href="mailto:klensin@jck.com">klensin@jck.com</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 Friday, 22 December, 2006 09:16 -0800 Mark Davis<br><<a href="mailto:mark.davis@icu-project.org">
mark.davis@icu-project.org</a>> wrote:<br><br>> What we say in PRI#96 is:<br>>> In each of the following contexts, the match to the regular<br>>> expressions<br>> must also only consist of characters from a single script
<br>> (after ignoring<br>> Common and Inherited Script characters).<br>><br>> While it does place limitations on fields containing joiner<br>> characters on<br>> the basis of script, it doesn't require the mixture of
<br>> scripts, in the sense<br>> used in<br>> <a href="http://www.unicode.org/reports/tr39/#Mixed_Script_Detection">http://www.unicode.org/reports/tr39/#Mixed_Script_Detection</a>.<br><br>I certainly understood that and did not intend to imply
<br>otherwise. What I was trying to say is that one of the<br>arguments against protocols rules prohibiting mixing of scripts<br>is that, with the exception of some bidi issues (which we got at<br>least partially wrong), IDNA2003 operates on characters, not
<br>complete labels. A mixed-script test requires making the step<br>into evaluating complete labels for correctness (under Michael's<br>proposal, complete FQDNs). That is a non-trivial step. I<br>believe that any sensible model for handling ZWJ and ZWNJ
<br>(including that of PRI#96, which I assume to be the default<br>unless better ways are found) will require looking at full<br>labels or at least sequences of characters, i.e., making that<br>step.<br><br> john<br><br>
</blockquote></div><br>