Hi Doug, Nick,
Please test this for me.  As I said, I don't get an error now, but I didn't before either.
I just ran a test here with the plugin that opens as 'r', and can't reproduce the error either.

So, does this mean that I was wrong about the 'r'? Is it my reflex from Windows times, where 'r' and 'rb' yield different results when reading text files? Is it that 'r' works fine on my test files, but one of Doug's files has a byte string that somehow triggers reading as UTF-8? Or could it be related to Doug not having GExiv2?

I suspect that opening as 'rb' won't hurt, but I do like to understand what's going on.