From: Jill Ramonsky (Jill.Ramonsky@Aculab.com)
Date: Tue Nov 11 2003 - 03:58:42 EST

• Next message: Don Osborn: "Re: Berber/Tifinagh"

Look, for one thing I mentioned natural sort of an EXAMPLE of how I
think the digits ten to fifteen should be treated identically to the
digits zero to nine, not the raison d'etre. But could anyone else who
wishes to post on this subject (natural sort) please CONSIDER whether
you've actually understood the subject before posting either strawman
arguments or just plain nonsense. The following are FACTS. As in,
mathematical facts. As in, I can prove them, and so can you, and we will
all reach identical conclusions provided that each step follows
logically from the previous ones. Note these then:

(1) A digit is a digit is a digit. There is no difference between a
digit is 5. This is the same digit which you find in radix-6, radix-93
or radix-7654321. Anyone who suggests there is ANY rationale for having
a separate set of digits for each radix is just plain wrong, and, I
would suggest, not a mathematician.

(2) The natural sort algorithm works identically in all radices. There
is nothing special about radix ten. Furthermore, the same sort order is
guaranteed in all radices. An implementation of a natural sort algorithm
does NOT need to "know" the radix. It does not need to guess. It does
not need to assume. It does not need to infer. It does not even need to
/care/. All it needs are the functions IsDigit(codepoint) and
GetDigitValue(codepoint). The return value of the latter is only
required to be defined if the return value of the former is true. That's
ALL it needs.

(3) Nobody in their right mind would even /consider/ that mixed radices
need to sort in numerical order. Such a thing is absurd. Jim Allen
(below) said "If you want a natural sort using a mixed alpha and numeric
string which may use multiple bases...". Well note the word "if" in that
sentence, because it is very pertinent. *I DON'T*. In fact, NOBODY on
this list has proposed that multiple radices should be intermixable. SO
WHAT if one thousand nine hundred and eleven (777 hex) is greater than
nine hundred and ninety nine? That does NOT mean that "File777" should
sort before "File999". In fact, Nobody, and I stress this because it's
starting to annoy me, *NOBODY* on this thread has supported the use of
mixed bases. Certainly not I. Therefore, anyone who argues against it,
is arguing against something which /nobody /has proposed. And that seems
about as good a definition of "waste of time" as I can think of.. It is
most certainly not an argument against anything I've said, as I am in
complete agreement with the notion that it's a totally dumb and stupid
idea. It's a strawman argument, and it's sidetracking away from the
original issue of _whether or not there should exist Unicode characters
for which IsDigit() returns true and for which GetDigitValue() returns
values in the range ten to fifteen_.

And finally, please note, Mr Allen, my name is Jill, not Jim. I think
you'll find that in fact _your_ name is Jim, though I can see why that
might be confusing. :-)

Jill

PS. This thread has been and interesting experience. If I do write a
letter of support, do I really have to go to such extremes to point out
what I'm NOT supporting? It would seem so.

> -----Original Message-----
> From: Jim Allan [mailto:jallan@smrtytrek.com]
> Sent: Monday, November 10, 2003 7:02 PM
> To: unicode@unicode.org
>
>
> Jim Ramonsky posted:
>
> > I am not the one who has not thought it through. There _is_ no
> > difference between decimal 7 and hex 7. They are the same
> digit. File777
> > sorts before File999 in _ALL_ radices.
>
> Exactly.
>
> So mixed hex and mixed decimal will not sort or compare
> properly using a
> natural sort *string* comparison even with creation of clones of the
> alpha characters with numeric values.
>
> Why then use a natural sort at all?
>
> If you want a natural sort using a mixed alpha and numeric
> string which
> may use multiple bases, a reasonable procedure might be

This archive was generated by hypermail 2.1.5 : Tue Nov 11 2003 - 04:59:58 EST