Deprecated characters?

John C Klensin klensin at
Thu Jul 17 23:21:23 CEST 2008

--On Thursday, 17 July, 2008 20:53 +0200 JFC Morfin
<jefsey at> wrote:

> Dear John,
> we obviously have different priorities you as an academic
> designer  and I as a pragmatic operator.

Gee, I haven't been accused of that in circa sixteen years years.

> However, I am sure
> you realise that if  a character is replaced by another
> because Unicode decides it, the  registered Trade Mark will
> follows the replacement if it is adopted  by ISO. Courts and
> TLD Managers will see it as a technical typo  correction
> having nothing to do with creating new IPRs (missing 
> clarifications about this kind of points is for them a worry).
> If we  are not sure of that, the best is then that we ask the

But here I have almost no idea what you are talking about.  The
characters that are proposed for deprecation are not, if I
understand any of Mark's comments, Ken's note, or the review
question as presented on the web site, "being replaced by other
characters".  Their use is simply being discouraged because they
should not have been assigned code points in the first place or
because there are better ways to do that same thing.  And, as
Ken's note points out, very few of them are allowed in IDNs
anyway (they are DISALLOWED, not PVALID, and hence irrelevant to
anything that could appear in a domain name).

ISO doesn't have anything to do with actions that affect
registered Trade Marks, and things that they "adopt" (in ISO/IEC
JTC1/SC2 or elsewhere) don't automatically change anything.

More generally, let's assume that there were a character that
Unicode and ISO 10646 decided declare obsolete, obsolete,
withdraw, and replace by something else -- that case is
theoretically possible but let me save Ken the time by saying
that it doesn't happen in practice and certainly is not what is
being discussed as "deprecation".  If I had a Trademark that
involved that code point and had reflected that Trademark in a
domain name, I would certainly rush out an register the name
name with the new spelling.  I would probably also retain the
old spelling forever, just to prevent various sorts of mischief.
And I would use the trademark laws against anyone who tried to
get ahead of me in registering the new form on the grounds that
the two forms were a lot closer together than what would
normally be considered similar enough to cause confusion.  And I
would do that regardless of whether the trademark itself somehow
magically migrated from one form to another.

And, because that whole issue affects decisions made by
trademark holders and registries, it has nothing to do with the
task list for this WG, so you are confusing policy issues for
which responsibility lies elsewhere with WG tasks.  Again.

> May be you misread my mail which was only to suggest a domain
> name  update to operators. This has nothing to do with the way
> the DNS  works nor any transitional effects which belongs to
> the Registry  Manager authority and competence. It was only
> that a reminder in the  single IDNA document that they should
> use would encourage them to use  and respect it; or it could
> be a mention in the security section?

I can't even parse the above paragraph in a way that would
permit me to take some action... if, in fact, you are proposing
that I take some action.

> The only thing is that as a Registry Manager I am not
> interested in  supporting deprecated chararcter domain names
> without them to be  duplicated through a PNAME. (Since I am
> prevented to document it  otherwise I call a PNAME a polynym
> [strict synonym in another  language/script] CNAME).

If, as a Registry Manager, you make a policy and operational
decision that every label that contains a deprecated character
should be associated with a CNAME (or DNAME) pointing to the
non-deprecated form, nothing prevents you from doing that.
Ken's analysis suggests that you are going to have that issue
only if you are Registry Manager who is registering names in
Khmer, but perhaps you do that.   Of course, if you tried to put
things into the DNS using an RR type that no one else has heard
of, you would encounter some operational difficulties.


More information about the Idna-update mailing list