On Sat, 5 Sep 2026 16:25:37 GMT, Alan Bateman <[email protected]> wrote:
>>> The head branch is benshuai5D:codex/JDK-8250677. Can you say if any part of
>>> the changes in this PR were generated by codex?
>>
>> No. AI was used only to help understand the surrounding code, identify
>> relevant existing tests, and organize test results.
>> The code changes in this PR were written and reviewed manually by me and my
>> colleagues.
>
>> I was used only to help understand the surrounding code, identify relevant
>> existing tests, and organize test results. The code changes in this PR were
>> written and reviewed manually by me and my colleagues.
>
> Thanks for confirming.
>
> I'm not sure yet about doing anything about this. The proposed change looks
> like it will break the contract for Object.hashCode in that two equals
> Version objects might have different hash codes, e.g. "1" and "1+0".
@AlanBateman
> I'm not sure yet about doing anything about this. The proposed change looks
> like it will break the contract for Object.hashCode in that two equals
> Version objects might have different hash codes, e.g. "1" and "1+0".
That’s already broken in the current version for `"1‑a"` and `"1‑a+0"`:
var a = ModuleDescriptor.Version.parse("1-a");
var b = ModuleDescriptor.Version.parse("1-a+0");
IO.println(a.equals(b)); // prints "true"
IO.println(a.hashCode() == b.hashCode()); // prints "false"
-------------
PR Comment: https://git.openjdk.org/jdk/pull/32715#issuecomment-5553375719
