71197 - Jis 208 may cause blank glyphs (FillInfo includes CP932)

Status: VERIFIED FIXED

(Core :: Internationalization, defect)

Description

[kill this account]

The jisx0208 converter FillInfo includes the Microsoft cp932 extensions.
When the moz X font is searching for a glyph these cp932 extensions
make it think there are glyphs when there are not.

This bug may be related to http://bugzilla.mozilla.org/show_bug.cgi?id=70146

Comment 1

[kill this account]

*** Bug 66744 has been marked as a duplicate of this bug. ***

Comment 2

[kill this account]

Attachment: revised converter file

Comment 3

[kill this account]

Attachment: diff between file used to generate the previous version and the proposed version of jis0208.uf

Comment 4

[kill this account]

At present the source files used to generate the *.uf/*.ut file are not
checked in. I have already opened a bug on this and associated this bug
with it: http://bugzilla.mozilla.org/show_bug.cgi?id=67374

The previous version of the file jis0208.uf was generated by running
http://lxr.mozilla.org/seamonkey/source/intl/uconv/tools/jis0208fromcp932.pl
on the CP932 table.

Comment 5

[kill this account]

Slight correction:

The previous version of the file file used to generate jis0208.uf was generated 
by running
http://lxr.mozilla.org/seamonkey/source/intl/uconv/tools/jis0208fromcp932.pl
on the CP932 table.

Comment 6

[nhottanscp]

r=nhotta

Comment 7

[Masaki Katakai]

I've tried your jis0208.uf and have verified the fix for the missing problems of

	bug 70146 (jisx0212)
	bug 66744 (GBK with gb2312 problem)

with disabling cns-4 entry in gfx/gtk. How is the result on your
envionment about bug 70146? Have you also verified? I belive you
can close bug 70146 as dup of this bug. Thanks.

Comment 8

[Masaki Katakai]

Once this problem gets fixed, I'll provide test binrary to our L10N team
(Brian@Sun) for evaluation. (with with disabling cns-4 entry in gfx/gtk).

Comment 9

[kill this account]

*** Bug 70146 has been marked as a duplicate of this bug. ***

Comment 10

[Erik van der Poel]

After some discussion with Brian and Frank, I am comfortable with the change.
sr=erik

Comment 11

[kill this account]

move this ot 0.9 since there is a discussion about reworking the converters

Comment 12

[kill this account]

this bug is related to bug 71762 as they currently share the
same jis 0208 converter

Comment 13

[Frank Tang]

OK, I will include two new patch so we will convert to JISx0208 in the standard
way and convert to EUC-JP and ISO-2022-JP with cp932 extension also. 

the jis0208.uf changes in the first patch should be the same as the one we
previous included.

we also need a new file jis0208excp932.uf in the same directory.

Comment 14

[Frank Tang]

Attachment: change to mozilla/intl/uconv/ucvja

Comment 15

[Frank Tang]

Attachment: new file which should be saved as mozilla\intl\uconv\ucvja\jis0208excp932.uf

Comment 16

[Frank Tang]

reassign to ftang

Comment 17

[Frank Tang]

need a new patch. We need to change the following from the current patch
1. the name of jis0208excp932.uf is missleading. Change to jis0208ext.uf
2. we should not generate escape sequence when we switch between jis0208.uf and 
jis0208ext.uf tables in Unicode ISO-2022-JP converters
3. We should not use C++ comment in .c file

Comment 18

[Frank Tang]

Attachment: new patch

Comment 19

[Frank Tang]

To fix this, we need to the following patch
1. the new jis0208.uf (see the 2nd attachment dated 03/08/01 13:59)
2. a new file called jis0208ext.uf (see the 4th attachment dated 03/14/01 17:03)
3. the latest patch (see the 5th attachment dated 03/28/01 16:13)

Comment 20

[Frank Tang]

Checked in everything that reviewed w/o no problem. Have not check in the 
nsUnicodeToISO20220JP.cpp change yet. Need some more time to test 

Comment 21

[Frank Tang]

mark this bug as fixed. we still need to fix 71762. 

Comment 22

[Yuying Long]

I'll mark it as verified due I didn't see those blank glyphs on 04-04 trunk build.