From: Philippe Verdy (firstname.lastname@example.org)
Date: Mon Nov 10 2003 - 08:44:09 EST
From: "Jill Ramonsky" <Jill.Ramonsky@Aculab.com>
> Microsoft Windows XP does a pretty good job of natural sort order. For
> example a file called "File99" will sort just before "File100".
> "File99A" will slot between them, but "File992" will go after them. It's
> all pretty much exactly what you'd expect.
> To sort "File10000" immediately after "FileFFFF" could be easily done
> using the a slightly modified form of same algorithm (maybe it would be
> a "tailored sort", I don't know), however the difficulty is in
> distinguishing the two meanings of 'F'. If it were clear that the first
> F was a letter and that the last four Fs were numeric digits with value
> fifteen, then such a sort would be a piece of cake to implement. So, the
> only important detail here is to distinguish number-F from letter-F. The
> rest is just an implementation problem.
So why do you vote for a proposal that simply does not address this issue?
What you want is a character with a numeric property and collated like
numbers, but the proposal addresses only the digit-figure constraint, and
in fact does not require anything for correct collation (and does not
justify as well the need for a numeric property for the proposed character,
which is viewed only as a presentation form but does not alter the
semantic of the new characters as letters...)
Creating constraints on new hex letter digits only is stupid in absence of
similar constraints on new decimal digits. Would we accept to change
the even more important legacy use of decimal digits to represent
numbers? Certainly not...
This archive was generated by hypermail 2.1.5 : Mon Nov 10 2003 - 09:25:16 EST