Re: [idn] proposed i18n naming rules

["James Seng/Personal" <jseng@pobox.org.sg>]

Yes, I meant U+002E, U+3002, U+FF0E and also U+FF61 (I missed that..:-)
should be considered to be defined as label separator.

Altho Nameprep has mention U+3002 (by prohibiting it), there is
inconsistency. e.g consider U+FF0E would actually normalized to U+002E
but U+3002 is prohibited.

And as Mark Andrew said, we should probably not prohibit the use of
these seperator in labels. It is up to the applications to determine to
detect the separator. Therefore, what we need to do is to actually
specific the lists of possible separator.

-James Seng

> James Seng/Personal wrote:
> > What we need to do is defined U+3002 as a DNS label seperator
> > *in addition* to U+002E. We should also consider U+FF2E to be
> > consistent.
>
> You mean U+FF0E (also U+FF61 if we map other halfwidth characters).
>
> There are two questions here:
>  - whether U+3002 and U+FF0E should be allowed as a separator in
>    application layer encodings of a domain name (no, absolutely not)
>  - whether user input processes should be allowed or encouraged to map
>    U+3002 and U+FF0E to U+002E before encoding them (yes, and nameprep
>    already says that).