xloem

Thanks for all your reviews. It looks like just a pretty small amount of work is needed to bring this up to community standards. If there are other people waiting...

What normanr is describing appears to be at https://github.com/green-green-avk/libusb-1.0.23-android-helper-service-patch/blob/master/libusb/os/linux_android_helper.c . It's unfortunate that alex initialized a new repository rather than forking libusb to track changes, but it is not too...

> For 3rd change, Is it mandatory to pass vm or just set LIBUSB_OPTION_ANDROID_JAVAVM in native like libusb_set_option(NULL, LIBUSB_OPTION_ANDROID_JAVAVM); You must pass the actual vm or jni pointer for the...

`PKG_CONFIG_EXECUTABLE=/bin/false` https://cmake.org/cmake/help/latest/module/FindPkgConfig.html#variable:PKG_CONFIG_EXECUTABLE

In a project I was working on, it was helpful to disable pkg-config when building for android. Setting it to /bin/false makes cmake recognise that it isn't usable. That way...

[sorry for extra spam] I think the OP is proposing adding a build step in the /android/ subfolder that would simplistically resolve all the issues. I imagine if a PR...

@mcuee 874 and these two PRs are presently orthogonal i.e. no it does not and I expect the changes should merge fine (but haven't looked recently).

[this comment previously documented some work I did towards implementing this. I have since done enough work that it's actually usable, and I'm moving its reference to a pull request.]

I'm happy to have this replace #874 or merge and rebase it too, whatever works better for maintainers and contributors.

My commit removing early string generation from my nonrooted android jni work is at 809a9180d53178dde812ebb7cfec8766bd74d3d5 if ever needed to make something easier and quicker.