Re: [idn] lowercased hostnames (was: naming syntax rules)

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

>> But for solutions using native UCS (in the form of UTF-8) in the DNS
>> protocol, we can allow DNS to work like today matching names
>> case-insensitively.
>
>Isn't this mutually-exclusive? If the hostname is defined with UCS
>characters of mixed case, then the PTR will have to always provide that

>case form, but ACE will not be able to provide the PTR data in mixed
case.
>Therefore, ACE will always break the PTR if a hostname is defined in
mixed
>case. And if people are allowed to define mixed case (the canonical UCS

>form works), then they will always be getting shafted by ACE PTR RRs.
>
>It seems that if we are forced to lowercase everything but ASCII for
ACE
>then we have to always lowercase everything but ASCII regardless of the

>encoding that may or may not be used.

No. It just shows that the IDNA solution will not always return the true

domain name. Instead it will return the form used for comparing
names using binary matching. It will be usable for all applications only
understanding
ACE and for all applications matching names using the rules we define
for that.

A solution using UCS for new software and ACE for backward compatibility

can use the true names in UCS as all applications understanding it is
required to know how to compare names. Old software will only see the
ACE version and will neve be aware of the true name.

    Dan