RE: [idn] Reality Check

["Brian W. Spolarich" <briansp@walid.com>]

| IDNA indirectely says the above!
| IDNA says that over the DNS protocol namepreped ACE must be used.
| Because of this, DNS can only return names in lower case.
| So, while you can enter mixed case, the moment a name is sent or
| received over the DNS protocol it is converted into lower case and
| the mixed case information is lost.

  This does not have to be the case.  An ACE encoder which takes case
information passed by nameprep can store case information for the decoded
name by setting case in the ACE sequence.  The decoder can then use these
hints when converting for display.

  This would require nameprep to capture case before squashing and passing
to an ACE which supports it.  Any of the ACEs which do not break codepoints
during their encoding steps could support this.  DUDE-02 does, as do all of
AMC's alternatives (I think).

  If the WG feels that case preservation in decoding is important, then 1)
it is not inconsistent with IDNA's architecture and 2) the ACE and NamePrep
can be modified to support it without adding too much complexity (beyond the
inherent complexities of case in general).

  -bws