[idn] Does IDN in applications only, solve the right problem?

[Dan Oscarsson <Dan.Oscarsson@trab.se>]

IDNA was made to allow non-ASCII in host names by making
only applications understand them.
And I have seen that some are experimenting on adding this
to web browsers.

But is this were it is needed?

The basic need of ACE is to allow ancient programs to still
being able to get a name in ASCII only.
But a web browser is a new program, it is easy to modify
and it could use UTF-8 directely.
On the other hand, the programs that ACE is there to protect, like
for example a lot of system software under Unix, many of them
are programs that really need to be fixed to handle non-ASCII.
Before I can tell my Unix host that its name contains non-ASCII
and have programs like inetd, logging, telnet and ftp handle
them, ACE have less need as browsers work fine with UTF-8.
Note: the programs above handle non-ASCII names when I can
enter the names into the programs and configuration files using
non-ASCII characters. The names may not be in ACE. Native encoding
is required.

So maybe the easiest thing to do is to rewrite the
resolver libraries to decode/encode UTF-8 or ACE into native
character set so that all programs never see ACE or UTF-8?

I am very doubtfull that ACE in application only, will
allow me to use non-ASCII in my host names soon, as it will
not fix many of the programs that most need to be fixed. Maybe
after 10 years.

   Dan