🚀 go-pugleaf

{}

I think we are in danger confusing two different aspects of sorting.

1/ The simple case of ``sorting'' in whatever-my-machine-likes order
   for table lookup (automated binary searches, hash lookup etc)

2/ Sorting for human consumption.  This is almost certainly not character
   set order, and may not be even remotely related (Yes even in English).

Type 1 is largely irrelevant to internationalisation, except in as much
as this is the type of operation carried out by *all* our
general-purpose utilities, but there is little need to change these, as
we are doubtless more interested in the internal efficiency than the
external order (cf. dbm).

The other question (type 2) is a much more involved operation.  I
suggest we all reach for D.E.Knuth's book ``The Art of Computer
Programming'' volume 3 ``Sorting and Searching'' pp 7-9 exercise 16.
This spells out the problem much better than I could (and cuts down
the total news traffic).

It is fairly obvious that real sorting for humans will involve
sufficient heuristics to make the natural order of the internal
character set immaterial.  It seems likely that such sorting will have
to be done on a per-language (and to a certain extent per-country)
basis.

This is not to say that multiple-alphabets and their internal representation
is not relevant and interesting, (and I can't contribute to that discussion)
merely that how such multi-lingual text sorts in simple per character
comparisons is almost a red herring.
-- 
Regards,
	Andrew Macpherson.	<andrew@stc.UUCP>
	{aivru,creed,datlog,iclbra,iclkid,idec,inset,root44,stl,ukc}!stc!andrew