On Tue, 13 Jan 2026 16:52:30 GMT, Phil Race <[email protected]> wrote:
>> Hi,
>>
>> new String(b, 0, n) looks at the operating system's default locale. If these
>> differ between the build machine and the test runner, tests can fail.
>> StandardCharsets.UTF_8 is guaranteed to be available on all Java platforms.
>>
>> Since "ERROR" and "Exception" are ASCII characters, and ASCII is a subset of
>> UTF-8, this will correctly match the tokens regardless of the underlying
>> platform's obscure defaults.
>>
>> Kindly review.
>>
>> Regards,
>> Anupam
>
>> If these differ between the build machine and the test runner, tests can
>> fail.
>
> How does the build machine locale matter ? Is something caching encoded
> output at compile time ?
Hi @prrace ,
No, there is no caching at compile time. I referred the test environment as
build machine for the class J2DdemoTest.
The test runner can be any IDE or different shell with ANSI. If J2Ddemo outputs
a special character (for example a non-standard character in a file path) that
doesn't map correctly in ANSI, the decoding might fail or garble the text
surrounding it.
So in spite of the error, the below two will never be true:
`if (test.output_contains("ERROR")) `
`if (test.output_contains("Exception")) `
Regards,
Anupam
-------------
PR Comment: https://git.openjdk.org/jdk/pull/29190#issuecomment-3747737943
