<div dir="ltr">One thing that I hope we have a chance to discuss in Dublin is interoperability. <br><br>IDNA2008 is actually much more lenient than IDNA2003, because it allows
arbitrary local mappings. Suppose you have any of the following in an email message,
for example.<br><ol><li>
<a href="http://schaffer.de">http://schaffer.de</a></li><li>
<a href="http://Schaffer.de">http://Schaffer.de</a></li><li>
<a href="http://Schäffer.de">http://Schäffer.de</a></li><li>
<a href="http://Schaeffer.de">http://Schaeffer.de</a></li><li><a href="http://Schæffer.de">http://Schæffer.de</a></li><li>
<a href="http://Schäffer.de">http://Schäffer.de</a> # using <<code><a target="c" href="http://unicode.org/cldr/utility/character.jsp?a=FB00">U+FB00</a></code> ( ff ) LATIN SMALL LIGATURE FF></li><li><a href="http://Schäf">http://Schäf</a><<code><a target="c" href="http://unicode.org/cldr/utility/character.jsp?a=00AD">U+00AD</a></code> SOFT HYPHEN><a href="http://fer.de">fer.de</a></li>
<li>
http://<<code><a target="c" href="http://unicode.org/cldr/utility/character.jsp?a=E0065">U+E0065</a></code> TAG LATIN SMALL LETTER E><<code><a target="c" href="http://unicode.org/cldr/utility/character.jsp?a=E006E">U+E006E</a></code> TAG LATIN SMALL LETTER N>Schäffer<<code><a target="c" href="http://unicode.org/cldr/utility/character.jsp?a=E007F">U+E007F</a></code> CANCEL TAG>.de</li>
</ol>(where <...> is a literal character)<br><br>An IDNA2008-conformant implementation could lowercase any of these using a local mapping -- or not, in which case #2-8 would fail. It could remove the illegal characters in #6 to #8, or not remove them and have the lookup fail. It could map the ligature ff to ff, or not. It could even decide, for example, based on locale linguistic mappings using the UI language of the client, or the language of the email, or the default system language, that it could map #3 to #2, #4 to #3 or vice versa, or #5 to #3.<br>
<br>On IDNA2003, in contrast, the mappings for all of these are completely determinant (with all but the first being allowed, and the last being disallowed). While implementations do do some prefiltering of certain format characters in some cases, except for that they tend to follow the rules. I've become more concerned over time that throwing the doors open to arbitrary mappings will end us up in an interoperability nightmare. See also the rough draft I had some time ago at <a href="http://docs.google.com/Doc?docid=dfqr8rd5_51c3nrskcx">http://docs.google.com/Doc?docid=dfqr8rd5_51c3nrskcx</a><br>
<br clear="all">Mark<br>
</div>