[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [idn] Forbidden characters
> >They are not forbidden: they will get folded to something
> else. So you
> >can still write and register your names with, say, U+00DC
> LATIN CAPITAL
> >LETTER U WITH DIAERESIS and U+2462 CIRCLED DIGIT THREE. They can be
> >used in your applications and by your users. They are just strictly
> >euqivalent to U+00FC LATIN SAMLL LETTER U WITH DIAERESIS and U+0031
> >DIGIT ONE. Just like 'A' and 'a' are strictly equivalent
> today in DNS.
> >
>
> But today the name service itself can handle both 'A' and
> 'a', and you can register a name with 'A' in it. When
> nameprep is required to be done before a name enters the name
> service, the name service cannot itself handle the above
> characters. So in the name service they are forbidden and
> cannot be returned from it. So for the DNS service (which
> includes the DNS servers and the protocol) the characters are
> forbidden.
Again, they are not forbidden. They can be used as input. What is not clear
(and controversial here) is whether you will ever get them back or not when
you decode the ACE. Some ACE preserve case. But normalization can't be
undone, so as you said, the name service cannot return them. We won't get
about adding a new RR for that here, but while I am for preserving case but
may live without it (after all, DNS names are not pretty, if I want
prettiness I'll use a layer above it), Roozbeh's comments about Arabic names
becoming different to their reader because of spontaneous joining of glyphs
certainly may spark some thinking.
YA