From: Mark E. Shoulson (mark@kli.org)
Date: Fri Feb 19 2010 - 11:06:19 CST
On 02/19/2010 06:56 AM, Kent Karlsson wrote:
>
> Den 2010-02-19 12.19, skrev "vanisaac@boil.afraid.org"
> <vanisaac@boil.afraid.org>:
>
>> Precomposed characters are currently discouraged from use,
>
> They are certainly not discouraged from use. Indeed, the preferred
> normalisation form, NFC, will use precomposed characters as much as
> possible (with a few exceptions...).
>
>> and I believe NFKD is the prefered normalization.
>
> NFKD is not at all a preferred normalisation. Indeed, it is only applicable
> in certain contexts, like IDN.
How could NFKx be the "preferred normalisation" in general?
Compatibility equivalence by design throws away information, and
information that is considered to be necessary for proper
representation. Things like GREEK PI SYMBOL were added precisely
because they were *not* used interchangeably with ordinary Greek PI by
some users. ALEF SYMBOL can't be interchanged casually with HEBREW
LETTER ALEF; one is LTR and the other is RTL. The entire MATHEMATICAL
series of alphabets in plane 1 is just about all
compatibility-equivalent to ASCII. Compatibility equivalence is there
to show resemblances and relationships, but not interchangeability
(except, as mentioned, in contexts where full expressiveness at the cost
of confusability is not desired)
~mark
This archive was generated by hypermail 2.1.5 : Fri Feb 19 2010 - 11:08:44 CST