L2/04-368

From: Mark Davis
Date: 2004-10-27
Subject: Uppercase and Lowercase properties

I propose that we change our properties so that Lowercase and Uppercase are
each subsets of Alphabetic. Currently there are 26 outlying values, circled
uppercase letters A-Z that are in Uppercase but not in Alphabetic

(a) This will allow for direct mapping to POSIX properties, and avoid
problems with conflicting names in Perl and other environments (ref proposed
draft of #18).

(b) I believe that most people interpret any property called Uppercase as
being Uppercase letters (ie Alphabetic), and thus expect the subsetting.

(c) The current definition of Uppercase/Lowercase is pretty wacky anyway.
Why include the circled uppercase letter symbols in Uppercase, but not
include some of the following symbols in Uppercase or Lowercase?

20A8           # Sc       RUPEE SIGN
2100..2101     # So   [2] ACCOUNT OF..ADDRESSED TO THE SUBJECT
2103           # So       DEGREE CELSIUS
2105..2106     # So   [2] CARE OF..CADA UNA
2109           # So       DEGREE FAHRENHEIT
2116           # So       NUMERO SIGN
2120..2122     # So   [3] SERVICE MARK..TRADE MARK SIGN
213B           # So       FACSIMILE SIGN
249C..24B5     # So  [26] PARENTHESIZED LATIN SMALL LETTER A..
				PARENTHESIZED LATIN SMALL LETTER Z
3250           # So       PARTNERSHIP SIGN
32CC..32CF     # So   [4] SQUARE HG..LIMITED LIABILITY SIGN
3371..337A     # So  [10] SQUARE HPA..SQUARE IU
3380..33DF     # So  [96] SQUARE PA AMPS..SQUARE A OVER M
33FF           # So       SQUARE GAL

(Some of these are 'mixed' case and so wouldn't belong for that reason;
I haven't bothered filtering them out. However, we don't have a Mixedcase
property right now; people have to use the Lt as an approximation, which
would exclude the above.)

And why not
U+24D0 CIRCLED LATIN SMALL LETTER A.. U+24E9 CIRCLED LATIN SMALL LETTER Z??

‎Mark