On Thu, Mar 10, 2011 at 20:58, Alan Cox <alan(a)lxorguk.ukuu.org.uk> wrote:
> The issue of what becomes what for purposes of comparison
> is easliy settled by asking a genealogist.
> Genealogists have been handling such problems for years.
More focussed on time shifting (eg รพ to th) but you could if you really
wanted.
You don't need to ask them however, librarians have been dealing with
case, sort order and the joys of "How do I file a book with a mixed
latin and greek title" long before the Columbus went yachting,
You can instead ask the standard. These days that is ISO Unicode 6.0
For transliterations see ISO Unicode LDR 1.9.0
For the joy of case conversion see Unicode Standard 6.0, UCB case
mappings and the supporting Annex (#31 if I remember rightly). Assuming
you've got through that and are not either lying on the floor gibbering
or down the pub attempting to forget what you saw (its a bit like Cthulhu
[1] it seems) you can continue to the implementation guidelines (about 30
pages of them), and read the glibc implementation thereof.
By which time you will most definitely not want to rely on caseless
comparisons as you will understand the true nature of caseless comparison
and how locale dependant it is.
I'm passingly familiar with the complexity involved. Nobody is asking
for the entire Unicode transliteration to be handled in Yum. In fact,
I don't know of any implementations at all.
The issue under discussion is the ASCII character set. No locale
treats case sensitivity for the ASCII-compatible first 128 codepoints
differently than the C locale. The feature requested would be
restricted to those codepoints.
--
Dotan Cohen
http://gibberish.co.il
http://what-is-what.com