> Yes, thanks, that patch fixes it so it builds successfully.  Tried w/
> Python 3.4rc1 on Windows 7 w/ VS2010.

Ok, good.

> Of course maybe it should really be building kill_python.exe for the
> matching architecture and running that, but I'm not sure how to do that
> and your fix is certainly far better than current behavior.

kill_python(_d).exe looks for a Python from the same directory as it is run from (with the same debug status) which should always mean that if kill_python(_d).exe can't run, neither can the Python it would look for.  My goal with this patch is to make running kill_python a "best effort" thing and make it obvious when it didn't run properly, without killing the build.  If it should have run but wasn't able to, there will be louder errors later on, and the warning from the attempt to run it should help point out what went wrong.