OS X doesn't support Valgrind or GProf. That's enough for me to keep a Linux VM around. Additionally, I've found that software is often easier to build under Linux.
It could definitely be a false positive. That said, that would be pretty strange behavior for a function called getColor(). I'd expect that to return a property value, not modify some state.
I can confirm that MonoDevelop works reasonably well for F#, but the documentation for F# on Mono isn't great. Coming from OCaml, I love F# as a language, but the tooling isn't as good as it is for OCaml. Might be better on Windows with VS, but I haven't looked closely at that.
That's pretty much what Nuitka is. Unfortunately, PyPy is only fast because it can take advantage of runtime information in the JIT. AOT compilation loses that advantage.
I think it still is. The output is a function fitting the data, so as long as that function is amenable to analysis, it doesn't really matter how it's found.
In fact, the paper discusses some modifications they made (reducing the number of parameters) to make the output kernels more intelligible to humans.
AFAIK, the original reason for the DRM was to prevent other companies from selling K-cups, so it would need a way to validate each cup, not just prevent reuse.
