maintainer-feedback requested: [Bug 236754] java/openjdk11: platform-specific include dir at wrong place

[bugzilla-noreply at freebsd.org]

Bugzilla Automation <bugzilla at FreeBSD.org> has asked freebsd-java mailing list
<java at FreeBSD.org> for maintainer-feedback:
Bug 236754: java/openjdk11: platform-specific include dir at wrong place
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=236754



--- Description ---
Based on a report on the mailing list.

Previously:

> [mosipov at mika-ion /usr/ports/java/openjdk8]$ tree
/usr/local/openjdk8/include/
> /usr/local/openjdk8/include/
> ├── classfile_constants.h
> ├── freebsd
> │   ├── jawt_md.h
> │   └── jni_md.h
> ├── jawt.h
> ├── jdwpTransport.h
> ├── jni.h
> ├── jvmti.h
> └── jvmticmlr.h

now

> [mosipov at mika-ion /usr/ports/java/openjdk8]$ tree
/usr/local/openjdk11/include/
> /usr/local/openjdk11/include/
> ├── bsd
> │   ├── jawt_md.h
> │   └── jni_md.h
> ├── classfile_constants.h
> ├── jawt.h
> ├── jdwpTransport.h
> ├── jni.h
> ├── jvmti.h
> └── jvmticmlr.h


freesbd vs bsd

in ports I see:
> [mosipov at mika-ion /usr/ports/java/openjdk8]$ find . -name jni_md.h
> ./work/openjdk/jdk/src/macosx/javavm/export/jni_md.h
> ./work/openjdk/jdk/src/windows/javavm/export/jni_md.h
> ./work/openjdk/jdk/src/solaris/javavm/export/jni_md.h
> ./work/openjdk/hotspot/src/share/vm/prims/jni_md.h

> [mosipov at mika-ion /usr/ports/java/openjdk11]$ find . -name jni_md.h
>
./work/openjdk-jdk11u-jdk-11.0.2-9-2/src/java.base/unix/native/include/jni_md.h
>
./work/openjdk-jdk11u-jdk-11.0.2-9-2/src/java.base/windows/native/include/jni_m
d.h
>
./work/openjdk-jdk11u-jdk-11.0.2-9-2/build/bsd-x86-normal-server-release/suppor
t/modules_include/java.base/bsd/jni_md.h
>
./work/openjdk-jdk11u-jdk-11.0.2-9-2/build/bsd-x86-normal-server-release/suppor
t/interim-image/include/bsd/jni_md.h
>
./work/openjdk-jdk11u-jdk-11.0.2-9-2/build/bsd-x86-normal-server-release/jdk/in
clude/bsd/jni_md.h
>
./work/openjdk-jdk11u-jdk-11.0.2-9-2/build/bsd-x86-normal-server-release/images
/jdk/include/bsd/jni_md.h

It seems to be a bug, because I expect this to be platform-specific, and not
generalized.

This also applies to openjdk12.