bug#23814: 24.5; bug of hz coding-system

[ynyaaa]

Eli Zaretskii <eliz@gnu.org> writes:

>> `encode-hz-region' uses `iso-2022-7bit' coding-system internally,
>> replacing it with the coding-system below will work.
>> 
>> (define-coding-system 'iso-2022-cn-gb
>>   "ISO 2022 based 7bit encoding only for Chinese GB2312."
>>   :coding-type 'iso-2022
>>   :mnemonic ?C
>>   :charset-list '(ascii chinese-gb2312)
>>   :designation [(ascii chinese-gb2312) nil nil nil]
>>   :flags '(ascii-at-eol ascii-at-cntl designation 7-bit safe)
>>   )
>
> What advantages does this change have?

`iso-2022-7bit' may encode same character to various strings,
while `iso-2022-cn-gb' encodes same charcter to same string.

(mapcar (lambda (cs) (encode-coding-string
                      (propertize "\x4e00" 'charset cs)
                      'iso-2022-7bit))
        '(chinese-gb2312 japanese-jisx0208 korean-ksc5601
                         chinese-cns11643-1))
=>("\e$AR;\e(B"
   "\e$B0l\e(B"
   "\e$(Cli\e(B"
   "\e$(GD!\e(B")

(mapcar (lambda (cs) (encode-coding-string
                      (propertize "\x4e00" 'charset cs)
                      'iso-2022-cn-gb))
        '(chinese-gb2312 japanese-jisx0208 korean-ksc5601
                         chinese-cns11643-1))
=>("\e$AR;\e(B"
   "\e$AR;\e(B"
   "\e$AR;\e(B"
   "\e$AR;\e(B")

`encode-hz-region' expects `chinese-gb2312' characters are encoded
with "\e$A" sequences, and replaces them to "~{".