Prohibiting mapping of PVALID characters

Kenneth Whistler kenw at
Thu Dec 10 01:17:26 CET 2009

Eric Brunner-Williams wrote:

> > Protocol, 5.2:
> > 
> > 5.2. Conversion to Unicode
> ...
> > 
> > The protocol doesn't state that the Label String Input
> > has to start out as NFC -- in fact the allowance for
> > character set conversion from other local character
> > sets into Unicode would actually preclude that.
> Whatever it is you are describing it isn't (yet) a label.
> It is a sequence of things, characters perhaps, and possibly a string, 
> but a label is what such a thing might become, not what it is ab initio.

I realize that. It is Section 5.1. "Label String Input" and
not yet a label.

But we are talking about a prohibition (a MUST NOT)
to be added in Section 5.2 about mapping PVALID
characters. When the very next sentence in Section 5.2
says that "The result MUST be a Unicode string in NFC form,"
and you can only get the "Label String Input" into
NFC form by (among other things) mapping PVALID
characters to other PVALID characters.

So what am I missing here? I am talking about the same
5.2 paragraph everybody else is, I hope. :-)


More information about the Idna-update mailing list