135850 - JOHAB converter : Hangul compatibility Jamo mapping is wrong

Status: VERIFIED FIXED

(Core :: Internationalization, defect)

Description

[Jungshik Shin]

The mapping of Hangul compatibility Jamo (at U+3130) has to be
treated separately from the rest of symbols in KS X 1001 for the Johab
converter. In the table on p.33 of KS C 5601-1992, you'll
find row 4 is different from row 2 and row6.

       EUC-KR code points   JOHAB code points
row 4:  A4D4-A4FE            DAD4-DAFE

Cmpare row 4 with adjacent rows (e.g. row 2, row 6) and you'll
find the pattern.

row 2:  A2A1-A2FE            D9A1-DAFE
row 6:  A6A1-A6FE            DBA1-DBFE

Hangul Jamos are excluded in row 4 because they have to be encoded
using '5-bit patterns' with Hangul filers. 

I'm sorry I forgot to mention this when Johab converter was implemented.
As I told about this Ken Lunde a few years ago and asked him to
change Adobe's Cmap table for JOHAB (he also mentions it in
p. 181 of his CJKV I.P.) I believe CMap table in
Solaris for Johab gets this right. That means, Hangul compatibility
Jamos are rendered 'empty' if ksc5601.1992-3(Johab) fonts
are selected.

Anyway, I'll make a quick patch (though not tonight :-) ) ...

Comment 1

[Teruko Kobayashi]

Comfirming.

Comment 2

[Jungshik Shin]

Attachment: a patch

Comment 3

[Jungshik Shin]

Attachment: Johab Jamo to Unicode mapping table

to use with attachment 78165 [details] [diff] [review]

Comment 4

[Jungshik Shin]

Attachment: Unicode to Johab Jamo mapping table

to use with attachment 78165 [details] [diff] [review]

Comment 5

[Jungshik Shin]

I put up a web page to reveal this bug at
http://jshin.net/moztest/compat_jamo.html
Unless Linux box is configured to have ksc5601.1992-3 fonts
(the easiest way to do is to use Adobe CID fonts)
this bug wouldn't manifest on Linux. Solaris has to be used
to reproduce this bug. 

Comment 6

[Jungshik Shin]

> As I told about this Ken Lunde a few years ago and asked him to
> change Adobe's Cmap table for JOHAB (he also mentions it in
> p. 181 of his CJKV I.P.) I believe CMap table in
> Solaris for Johab gets this right. 

  According to Brian Yuan @ Sun (who added himself to CC), Sun
shipped CMap files obtained from Adobe without any change. Therefore,
what I wrote above is confirmed and this bug clearly affects Solaris. 

Comment 7

[Jungshik Shin]

Frank and Roy,
Could you review my patch?
Without this patch, under Solaris(and Linux and other Unix/X11
platforms) Hangul Jamos wouldn't be rendered
correctly if ksc5601.1992-3 fonts are used. 

Comment 8

[Roy Yokoyama]

Comment on attachment 78165 [details] [diff] [review]
a patch 

/r=yokoyama

Comment 9

[Roy Yokoyama]

Comment on attachment 78166 [details]
Johab Jamo to Unicode mapping table

/r=yokoyama

Comment 10

[Roy Yokoyama]

Comment on attachment 78167 [details]
Unicode to Johab Jamo mapping table

/r=yokoyama

Comment 11

[Alec Flett]

Comment on attachment 78165 [details] [diff] [review]
a patch 

sr=alecf

Comment 12

[Alec Flett]

maintenance on this first patch would be easier if you made better use of sizeof():

-: nsMultiTableDecoderSupport(4,
+: nsMultiTableDecoderSupport(5,

this could be
+: nsMultiTableDecoderSupport(sizeof(g_JOHABRanges) / sizeof(g_JOHABRanges[0]),

which will calculate the correct size at compile-time.

same thing for nsMultiTableEncoderSupport, appearing twice later in the patch.

Comment 13

[Alec Flett]

Comment on attachment 78166 [details]
Johab Jamo to Unicode mapping table

sr=alecf

Comment 14

[Alec Flett]

Comment on attachment 78167 [details]
Unicode to Johab Jamo mapping table

sr=alecf

Comment 15

[Jungshik Shin]

Attachment: a new patch incorporating alecf's comment

I'm going to ask for approval of drivers@mozilla.

Comment 16

[Alec Flett]

Comment on attachment 91119 [details] [diff] [review]
a new patch incorporating alecf's comment

sr=alecf

Comment 17

[Asa Dotzler [:asa]]

Comment on attachment 78166 [details]
Johab Jamo to Unicode mapping table

a=asa (on behalf of drivers) for checkin to the 1.1 trunk.

Comment 18

[Asa Dotzler [:asa]]

Comment on attachment 78167 [details]
Unicode to Johab Jamo mapping table

a=asa (on behalf of drivers) for checkin to the 1.1 trunk.

Comment 19

[Asa Dotzler [:asa]]

Comment on attachment 91119 [details] [diff] [review]
a new patch incorporating alecf's comment

a=asa (on behalf of drivers) for checkin to the 1.1 trunk.

Comment 20

[Jungshik Shin]

checked in to the trunk.
thank you for all.

Comment 21

[Teruko Kobayashi]

Jungshik Shin, could you create the test case for this, so I can verify this.

Comment 22

[Teruko Kobayashi]

Verified.