On Thu, 12 Dec 2024 16:00:52 GMT, Jorn Vernee <[email protected]> wrote:
>> According to above's prior-art, it could just read `release of {0} too high:
>> {1}`
>
>> `release of {0} too high: {1}`
>
> I think this is better. Probably good enough.
>
> My concern stems from the fact that 'release' doesn't seem to be defined for
> _class files_. I see that the term 'release' exists, and maps to a versioned
> directory, or you can compile classes _for_ a certain release, But, it seems
> that 'the release of a class file' is a new concept here. (after all, `javac`
> doesn't produce class files _with_ a certain release).
>
> Maybe: `class file {fp.entryName()} compiled for release
> {fp.classReleaseVersion()}, is illegal in version directory {fp.mrversion()}`
>
> (This is perhaps somewhat terse, but `javac` sets the precedent for that. A
> user could find out _why_ the version is illegal by referencing the jar file
> spec).
Let's try it out with the two examples from the initial description:
release version of META-INF/versions/9/version/Version.class too high: 25
release version of META-INF/versions/10/version/Version.class too high: 25
class file META-INF/versions/9/version/Version.class compiled for release 25,
is illegal in version directory 9
class file META-INF/versions/10/version/Version.class compiled for release 25,
is illegal in version directory 10
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/22103#discussion_r1882481827
