L2/01-189

Updates to EastAsianWidth   

Date: April-30-2001
From: Asmus Freytag

A review of East Asian legacy sets found
the following characters occuring as wide
characters in EA legacy usage. Their East
Asian Width should therefore be adjusted
to "A".

00AE REGISTERED SIGN
00AF MACRON (this could be a mistake in mapping the legacy set)
014B LATIN SMALL LETTER ENG
02C4 MODIFIER LETTER UP ARROWHEAD
02DF MODIFIER LETTER CROSS ACCENT
2022 BULLET
2024 ONE DOT LEADER
203E OVERLINE
2116 NUMERO SIGN
2153 VULGAR FRACTION ONE THIRD
215C VULGAR FRACTION THREE EIGHTHS
215D VULGAR FRACTION FIVE EIGHTHS
21B8 NORTHEAST ARROW TO LONG BAR
21B9 LEFTWARDS ARROW TO BAR OVER RIGHTWARDS ARROW TO BAR
21E7 UPWARDS WHITE ARROW
273D HEAVY TEARDROP-SPOKED ASTERISK

The Korean Johab set (Windows code page 1361)
shows all Jamo in the 11xx range, including the HCF,
only some of which are currently wide. These should
all be changed to "W".

The following list the characters by location in 
the EA legacy sets investigated.

20000 CNS
Eten and TCA have the same characters but in different locations - so the set is consistent for Taiwan)
A1A5 = 2024 - ONE DOT LEADER
A1A6 = 2022 - BULLET
A2A3 = 203E - OVERLINE
A3D0 = 02C4 - MODIFIER LETTER UP ARROWHEAD
A3DD = 273D - HEAVY TEARDROP-SPOKED ASTERISK
A5ED = 02DF - MODIFIER LETTER CROSS ACCENT
C2E7 = 2116 - NUMERO SIGN
FEEA = 21E7 - UPWARDS WHITE ARROW
FEEB = 21B8 - NORTHEAST ARROW TO LONG BAR
FEEC = 21B9 - LEFTWARDS ARROW TO BAR OVER RIGHTWARDS ARROW TO BAR

odd mappings
FEF0 = FF78 - HALFWIDTH KATAKANA LETTER KU

Big 5
A1C2 = 00AF - MACRON
(potentially a mismapping for 203E as 203E is not mapped from this set)

936 Simplified Chinese
A1ED = 2116 NUMERO SIGN

932 Japanese
8782 = 2116 NUMERO SIGN
FA59 = 2116 NUMERO SIGN

1361 Korean Johab
DD3F = 014B LATIN SMALL LETTER ENG
DCF7 = 2153 VULGAR FRACTION ONE THIRD
DCFC = 215C VULGAR FRACTION THREE EIGHTHS
DCFD = 215D VULGAR FRACTION FIVE EIGHTHS
D9E0 = 2116 - NUMERO SIGN
D9E7 = 00AE - REGISTERED SIGN
plus all 11xx Jamo not yet "W"
(Code page 949 agrees, except it doesn't have the Jamo)

Note on mappings: All of the sets investigated map wide ASCII 
and Latin-1 characters to the FULL WIDTH BLOCK. This is 
different from the sample mapping tables on the Unicode
FTP site. However, those files have not been maintained
since 1994, predate Unicode 2.0 and should probably be
adjusted.