Re: [idn] Debunking the ACE myth

["Adam M. Costello" <amc@cs.berkeley.edu>]

"D. J. Bernstein" <djb@cr.yp.to> wrote:

> This is an ACE failure...

In your scenario you paste a UTF-8 string into an IDN-unaware
application and it doesn't work.  I don't see how this is an ACE
failure.  It has nothing to do with ACE.

> These failures can't occur when there are no ACE applications.

False.  If the application is IDN-unaware but accepts 8-bit domain
names, then it will be able to look up the UTF-8 name if and only if the
old resolver API accepts UTF-8 queries.  We can debate whether it should
or not, but that issue is orthogonal to the issue of whether ACE exists
or not.

I think we all understand that ACE can serve to bridge IDN-aware and
IDN-unaware applications only when the IDN-aware applications know
that they have to use ACE (either because they know they're talking
to an IDN-unaware peer, or because they're talking via an IDN-unaware
protocol).  When the domain names are transferred out-of-band (via
cut-and-paste, for example), then the ACE-encoding might not happen.

Users who transfer domain names out-of-band will have to learn that
feeding UTF-8 to old applications is asking for trouble, and that's
equally true whether ACE exists or not.

I think there has been enough discussion to establish who favors ACE and
who doesn't, and you have no chance of dissuading the ACE supporters.  I
suggest that your time would be better spent working on UTF-8 standards
than on trying to prevent ACE standards.

AMC