Re: [CSS21] Case-insensitivity not defined

Addison Phillips scripsit:

> Interestingly, this question came up in my review of XmlHttpRequest just 
> yesterday. I believe that what you want is:
> 
> - You want to define it in terms of the Unicode definition.
> 
> - You also probably want to define it in deterministic terms, rather 
> than allowing it to be language sensitive. This means *not* using 
> SpecialCasing.txt or language-specific tailorings (e.g. the 
> Turkish/Azerbaijani dotted/dotless i mappings).

This is about case *folding*, not about changing cases, so neither
UnicodeData nor SpecialCasing is the right file; instead you want to
use CaseFolding.

CaseFolding provides two orthogonal options:

a) use the non-length-preserving foldings or not

b) use the language-sensitive folding or not

I agree that using the language-sensitive foldings makes no sense.
However, I see no reason to confine ourselves to the length-preserving
foldings, which are for when something is being folded "in place"
and it's critical not to use more Unicode characters than before.

-- 
The man that wanders far                        cowan@ccil.org
from the walking tree                           http://www.ccil.org/~cowan
        --first line of a non-existent poem by:         John Cowan