http://bugs.winehq.org/show_bug.cgi?id=36512
--- Comment #4 from Qian Hong fracting@gmail.com --- (In reply to Vincent Povirk from comment #3)
I think testing a nonexistent file is good enough to support patch 2. If one kind of failure sets the result to NULL, we can assume they all do.
When you put so much effort into testing different kinds of errors, it makes your tests harder to review, instead of helping support the change you want to make.
Thanks Vincent, patch resent: http://www.winehq.org/pipermail/wine-patches/2014-July/133448.html