Re: [idn] Report from the ACE design team

[Makoto Ishisone]

In message <[EMAIL PROTECTED]>,
Eric Brunner-Williams in Portland Maine <[EMAIL PROTECTED]> wrote:
> Can you comment on DUDE and smaller
> sets of characters with this distribution property (two or more "clusters",
> some low-range, some mid-range and/or some private-use range)?

DUDE uses a kind of differential encoding -- that is, the encoding
length of a single character is solely dependent on the "difference"
between the code point of the character and the previous one. If the
difference is confined in their lower 4 bits, the encoding length is
one.  If it is confined in the lower 8 bits, the length is two, and so
on.

So, even if there are two clusters, as long as the switching between
clusters is not frequent, DUDE's compression algorithm is effective.

Also DUDE's algorithm (and other ACEs as well) is independent of
whether a character is in a private-use area or not.  However, I think
private-use characters are prohibited by NAMEPREP and cannot be used..

                                              -- [EMAIL PROTECTED]