[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Comparison of hoffman-idn-reg and jseng-idn-admin

Roozbeh Pournader <roozbeh@xxxxxxxxxx> wrote:

> 1. Mandatory equivalences as opposed to secondary/variant
> equivalences.  This feature is necessary for defining equivalences
> between European and Arabic-Indic digit shapes in Arabic labels, for
> example.
> This is some feature that *all* Arabic script zones *require*

I'm not sure what you mean by "mandatory" and "require".

Suppose I create the names <ArabicTextX>.nicemice.net and
<ArabicTextY>.nicemice.net on the DNS server for nicemice.net (which I
control), and suppose these two names are the same except that one uses
ASCII digits and the other uses the corresponding Arabic-Indic digits,
and suppose I associate different resource records with these names.

Should there be a standard that prohibits me from doing this?

Or suppose I register <ArabicTextX>.com, which allocates an entire
bundle of names (including <ArabicTextY>.com) that no one else
can register.  Should the registry for .com be able to offer me
the choice of whether <ArabicTextY>.com is visible (equivalent to
<ArabicTextX>.com) or invisible (no such domain)?  Or should the
registry for .com be required to make <ArabicTextY>.com visible and
equivalent to <ArabicTextX>.com, regardless of my preference?

What would you think of this model:  The Arabic-speaking community
develops a best-practice recommendation regarding equivalences of
names that should resolve to the same resource records, and TLDs can
opt to support those equivalences, and can advertise their voluntary
conformance in order to attract Arabic registrants.

By the way, I'd like to point out that trying to make names
automagically appear to be equivalent is a much trickier task than
merely preventing registration of too-similar names to different
registrants.  The latter task involves the registry database only, not
the DNS servers, nor any other servers or applications.  But automagic
equivalence potentially depends on the cooperation of anything that
compares domain names.  For example, HTTP servers compare the requested
host name with host names in their config files in order to decide
which virtual server to present, web browsers compare host names with
user-entered lists to decide whether to accept images and cookies from a
given site, and mail servers compare recipient domain names with domain
names in their config files in order to decide whether the mail is

It might be easier to handle this sort of thing more manually, at least
in the beginning.  For example, if <ArabicTextX> contains four digits,
it's one of 16 equivalent names.  But really only two of them have any
chance of being used.  A registrant could select those two names to
be visible, and leave all others invisible.  They could then manually
configure their web server and their mail server to recognize those two

> 2. Clear language about conflict resolution.  There needs to be
> some clear guidelines or recommendations about the times that two
> registered labels come into an intersection regarding the variant
> labels associated to them.

The whole point of having bundles is so that two registered labels
cannot be variants of each other.  Whichever one was registered first
blocks the other from being registered.

Now this does suggest a new sort of problem that hasn't existed before:
bundles that are too big.  Currently, the problem is that bundles are
too small (just one name), so that registering a name doesn't prevent
someone else from registering a confusingly similar name.  The opposite
problem would be bundles that are too big, where registering one name
prevents someone else from registering another name that's different
enough to be considered distinct under trademark law (or whatever rules
are relevant to the dispute).  It might happen that registrant X has a
trademark on nameX, and registrant Y has a trademark on nameY, so both
registrants are equally "entitled" to their respective names, but the
two names are in the same bundle, so only one of the registrants can
have their name.  I guess this is not so different from cases where two
companies have trademarks on the same name in different markets, but
only one of them can have the name in .com.