<i18n dev> RFR: 8299811: SHIFT_JIS compatible charsets should be in java.base module on Linux platform

[Ichiroh Takiguchi]

On Mon, 9 Jan 2023 22:20:57 GMT, Naoto Sato <naoto at openjdk.org> wrote:

>> On Japanese environment, following SHIFT_JIS compatible charsets are there:
>> * PCK (x-PCK)
>> * MS932 (windows-31j)
>> * IBM943C (x-IBM943C)
>> 
>> According to make/data/charsetmapping/charsets,
>> PCK and MS932 are independent charset and have no dependencies on others.
>> 
>> On RHEL8, SHIFT_JIS encoding is supported
>> 
>> $ env LANG=ja_JP.sjis locale charmap
>> SHIFT_JIS
>> 
>> According to make/data/charsetmapping/stdcs-linux,
>> SJIS (SHIFT_JIS) is in there, also MS932 and PCK are there.
>> This means these are in java.base module on Linux platform.
>> But IBM943C is not in stdcs-linux.
>> I think IBM943C also should be in java.base module because of compatibility.
>
> I think the general consensus is to keep the `java.base` module minimal. IIRC, the reason we have `PCK` charsets in Un*x platforms was that it was required to boot Solaris in Japanese env (sorry if I am wrong). So I would remove it from `java.base` if possible. Thus moving those IBM* encodings that are equivalent to SJIS into `java.base` is something I would not want to do. It is definitely not a compatibility issue.

Hello @naotoj .

Sorry, "compatibility" may not be valid wording.
Now Linux is one of native application migration platform from *nix platform.
Since SJIS charset does not have enough supported characters on Japanese environment.
PCK and Cp943C charsets may be covered on this situation.

-------------

PR: https://git.openjdk.org/jdk/pull/11908