RE: IDN inclusion-based model

From: Phillips, Addison (addison@amazon.com)
Date: Wed Jul 23 2008 - 16:29:34 CDT

  • Next message: Erkki I. Kolehmainen: "RE: Court rules against Yup'ik as an historically written language (fwd media link)"

    I suppose the question would be: valid how? Nameprep (RFC 3491) is narrower than just "any Unicode code point". That is, IDNA imposes some restrictions on the range of valid code points. Certainly a user-agent has to deal with the possibility of "any" Unicode code point being entered into an address bar, even those illegal in IDNA addresses. And note that many of those characters *are* valid elsewhere in an IRI (so you can't just expunge them).

    Addison

    Addison Phillips
    Globalization Architect -- Lab126
    Chair -- W3C Internationalization WG

    Internationalization is not a feature.
    It is an architecture.

    > -----Original Message-----
    > From: unicode-bounce@unicode.org [mailto:unicode-bounce@unicode.org]
    > On Behalf Of Chris Weber (Casaba Security)
    > Sent: Wednesday, July 23, 2008 1:47 PM
    > To: 'Stephane Bortzmeyer'
    > Cc: unicode@unicode.org
    > Subject: RE: IDN inclusion-based model
    >
    > So then is it safe to assume IDN Character Categorization
    > documented at
    > http://www.unicode.org/reports/tr36/idn-chars.html are
    > recommendations only?
    >
    > Further is it safe to assume that user-agents like web browsers are
    > free to
    > recognize any Unicode code point (even the IDN-illegal ones from
    > above) as
    > valid?
    >
    > Chris
    >
    >
    > -----Original Message-----
    > From: Stephane Bortzmeyer [mailto:bortzmeyer@nic.fr]
    > Sent: Wednesday, July 23, 2008 1:23 AM
    > To: Chris Weber (Casaba Security)
    > Cc: unicode@unicode.org
    > Subject: Re: IDN inclusion-based model
    >
    > On Wed, Jul 23, 2008 at 12:33:06AM -0700,
    > Chris Weber (Casaba Security) <chris@casabasecurity.com> wrote
    > a message of 129 lines which said:
    >
    > > Is the decision for what's allowed to be included up to the
    > > registrars?
    >
    > In most of the cases I know, it it up to the registry.
    >
    > > At that point, should it be up to user-agents or registrars, or
    > > both, to ensure that legality is checked?
    >
    > All the registries I know check that the requested domain name
    > complies with their local rules which include the list of
    > authorized
    > characters.
    >
    > A registry cannot rely on users or registrars to do the check
    > because
    > not all of them are nice people.
    >
    >
    >
    > From: unicode-bounce@unicode.org [mailto:unicode-bounce@unicode.org]
    > On
    > Behalf Of Chris Weber (Casaba Security)
    > Sent: Wednesday, July 23, 2008 12:33 AM
    > To: unicode@unicode.org
    > Subject: IDN inclusion-based model
    >
    > How does the IDN Character Categorization documented at
    > http://www.unicode.org/reports/tr36/idn-chars.html fit into the
    > 'inclusion-based approach' referenced in
    > http://www.unicode.org/reports/tr36/#Country_Specific_IDN_Restricti
    > ons,
    > ICANN guidelines, and the RFC 4690?
    >
    > Is the decision for what's allowed to be included up to the
    > registrars? Or
    > can I gather from the idn-chars.html above that U+FF0F (FULLWIDTH
    > SOLIDUS),
    > for example, is an illegal? At that point, should it be up to
    > user-agents
    > or registrars, or both, to ensure that legality is checked?
    >
    > Thank you,
    > Chris
    >

    This archive was generated by hypermail 2.1.5 : Wed Jul 23 2008 - 16:31:23 CDT