Hm, seems I have some more work to do here. I'm not sure about the best strategy for threadsafety by default in the RandomAccessFile::ReadAt functions. Before there was a default virtual implementation, but the mutex was visible in the header. We could add a PIMPL to RandomAccessFile but that might be overkill.

The default option is that I add basic implementations to the classes that are breaking the build currently

This should be ready to go once the tests pass

hm, there's a test failure, it looks like I borked something with threadsafety. taking another look

@kou it's possible that this patch introduces threadsafety issues in the GIO wrapper implementations if you use the RandomAccessFile::ReadAt functions at all, just as a heads up

Thanks for the heads-up.
I'll add thread-safe ReadAt implementation to garrow::GIOInputStream later.

Waiting for https://ci.appveyor.com/project/wesm/arrow/build/1.0.1269 since I twiddled some symbol visibility stuff

+1