Re: [idn] Reality Check

[Keith Moore <moore@cs.utk.edu>]

> > first we need to get an ASCII-based solution out the door.
> 
> Not "out the door" so much, but one definitely needs to exists as a legacy
> mode. IOW, ACE must exist in any event, but it does not need to exist as
> "the solution to the problem".
> 
> > perhaps by then someone will have come up with a convincing argument
> > as to why having the ability to use UTF-8 on the wire in the DNS
> > protocol (and producing consistent results with ASCII queries)
> > actually produces a tangible benefit that is worth the incremental
> > cost.
> > 
> > so far, nobody has managed to do that.
> 
> Nobody has proposed a protocol for it. I will do this soon.
> 
> The costs for embedding and managing encrypted domain names in protocols
> is obviously greater than the costs for embedding and managing UTF8 domain
> names. This should be so obvious that no "arguments" are needed.

if you think the cost savings are obvious, you haven't done the analysis.

hint: the cost of dealing with two different name representations is greater
than the cost of dealing with a single one.

Keith