**From:** Jill Ramonsky (*Jill.Ramonsky@Aculab.com*)

**Date:** Mon Nov 10 2003 - 10:49:00 EST

**Previous message:**Alexander Savenkov: "Re[2]: Berber/Tifinagh (was: Swahili & Banthu)"**Maybe in reply to:**Michael Everson: "Hexadecimal digits?"**Next in thread:**Chris Jacobs: "Re: Hexadecimal digits?"**Reply:**Chris Jacobs: "Re: Hexadecimal digits?"**Reply:**Kent Karlsson: "RE: Hexadecimal digits?"**Messages sorted by:**[ date ] [ thread ] [ subject ] [ author ] [ attachment ]**Mail actions:**[ respond to this message ] [ mail a new topic ]

Sorry, but I have to correct you. You state below that "[my] argument

doesn't work". This is slightly confusing because I haven't proposed any

arguments, beyond that I support the inclusion into Unicode of hex

digits which are distinct from the letters A to Z.

I can only assume you are suggesting that the "natural sort" algorithm

works in decimal but not in hex. If so, I should mention that (1) it

wasn't me who invented the natural sort algorithm, so I can't take

credit for that in /any/ radix, and (2) there is absolutely no reason

why it wouldn't work in radix sixteen just as it would in radix ten. For

example 77 and 100 get sorted in the order (77, 100), not (100, 77) in

EVERY radix in which the digits 0, 1 and 7 exist. This is true in base

eight. It is true in base ten. It is true in base sixteen. It is even

true in base 68431. In fact, the only things the natural sort algorithim

need to know are (a) which characters represent digits and which ones

don't, and (b) what is the numerical value of each such digit.

To clarify this, if we applied the natural sort algorithm to the

filenames thus far mentioned, using a natural sort algorithm which

recognised the codepoints U+218A to U+218F as having the digit property

with values ten to fifteen respectively, then they would sort as

follows. (The algorithm wouldn't specifically need to know that the

digits were "hex", only that they were numeric).

(1) U+46, U+69, U+6C, u+65, U+39, U+39 ("File99")

(2) U+46, U+69, U+6C, u+65, U+39, U+39, U+41 ("File99A") -- note that

this A is a letter

(3) U+46, U+69, U+6C, u+65, U+39, U+39, U+46 ("File99A") -- note that

this F is a letter

(4) U+46, U+69, U+6C, u+65, U+31, U+30, U+30 ("File100")

(5) U+46, U+69, U+6C, u+65, U+39, U+39, U+32 ("File992")

(6) U+46, U+69, U+6C, u+65, U+39, U+39, U+218A ("File99A") -- note that

this A is not a letter, but the digit ten

(7) U+46, U+69, U+6C, u+65, U+39, U+39, U+218F ("File99F") -- note that

this F is not a letter, but the digit fifteen

If you are suggesting that the natural sort algorithm won't work

/without/ separate codepoints for hex digits then you are of course

correct, but that is an argument in favor of hex-digit-characters, not

against them.

Jill

* > -----Original Message-----
*

* > From: Kent Karlsson [mailto:kentk@cs.chalmers.se]
*

* > Sent: Monday, November 10, 2003 2:42 PM
*

* > To: unicode@unicode.org
*

* > Subject: RE: Hexadecimal digits?
*

* >
*

* >
*

* > > After all, 99A (in hexadecimal) is greater than 99 (hexadecimal).
*

* >
*

* > Oops. I missed the "2" key. E.g:
*

* >
*

* > > After all, 99A (in hexadecimal) is greater than 992 (hexadecimal).
*

* >
*

* > Sorry (both about missing the "2" and that your argument doesn't
*

* > work)
*

* > /kent k
*

* >
*

**Next message:**Jill Ramonsky: "RE: Hexadecimal digits?"**Previous message:**Alexander Savenkov: "Re[2]: Berber/Tifinagh (was: Swahili & Banthu)"**Maybe in reply to:**Michael Everson: "Hexadecimal digits?"**Next in thread:**Chris Jacobs: "Re: Hexadecimal digits?"**Reply:**Chris Jacobs: "Re: Hexadecimal digits?"**Reply:**Kent Karlsson: "RE: Hexadecimal digits?"**Messages sorted by:**[ date ] [ thread ] [ subject ] [ author ] [ attachment ]**Mail actions:**[ respond to this message ] [ mail a new topic ]

*
This archive was generated by hypermail 2.1.5
: Mon Nov 10 2003 - 11:34:07 EST
*