Re: [idn] report of the straw poll

["D. J. Bernstein" <djb@cr.yp.to>]

Paul Hoffman / IMC writes:
> a vast smattering of other choices with none having even 10% of the support.

In fact, several of those ``other choices'' were UTF-8; some of us
support UTF-8 but didn't list a choice in the poll; and we don't know
how many of the IDNA supporters also support UTF-8.

Bottom line: This poll was not presented to us as a measurement of UTF-8
support. It is thoroughly inappropriate for you to be using it that way.

> nameprep, which at this point is finished

No, it isn't, because it allows Beta, in violation of the nameprep-02
requirement that normalization ``must not introduce any chance that two
host names could be represented by two distinct strings of characters
that look identical to typical users.''

You seem to have removed that requirement without WG approval, but the
problem remains: you are allowing B and Beta in host names, normalized
to the distinct characters b and beta, even though B and Beta have
identical appearances in print.

---Dan