@bhamail you handled the last license adjustment, could you have a look at this please? Basicly every source file (outside libffi source) either gained the LGPL2.1+AL2.0 header or the existing header was adjusted to match that.

@twall Could you please have a look at the only functional change: I replaced the copied WeakIdentityHashMap with WeakMemoryHolder, that is a minimal implementation of the used function set.

The primary concern of #252 is addressed by this, the two secondary points:

• dlmalloc.c is PD, so I'd not change current handling (in all fairness I'm not sure what the point should be)
• X.java claims it is base on - not derived from wmctrl - while you can see similarities this is not a copy

One thing I just noticed - the file headers I see mention LGPL 2.1 or (at your opinion) any later version..

So my question: Is the reduction to LGPL2.1 intended? If not I'll rework this to be reflected.

@bhamail @dblock @twall
I'd like you to give your feedback on this PR. I was not around when the license change happened, so I can't be sure what was intended.

Summary:
It was decided, that the project should also be licensed under the Apache License 2.0. This is reflected by the README.md from the root directory of the source code.

This is in contrast to LICENSE, which only speaks about LGPL 2.1 and AL 2.0, but drops the "or later" part of the LGPL wording in README.md.

The source code files then contradict this by one being licensed under LGPL 2.1 or later.

In addition source code that is licensed under pure AL 2.0 was copied into the tree (WeakIdentityHashMap).

So to clear this up:

• Headers of Java/C sources were adjusted to reflect to AL2.0/LGPL2.1 (or later) duallicensing
• WeakIdentityHashMap was replaced with a custom minimal replacement implementation
• Contribution.md was update to reflect the licensing

This looks acceptable to me.

I vote 👍 too.

@twall I think this is your call to hit merge

I merged the changes into master - thank you for checking and chiming in.