Moving this review onto Phab.
Thanks for your comments Chris. I have added some tests for this feature and the original cflag feature, for which I didn't find existing tests.
To test the command-line argument, I added a bit of hack in nt.py which just add a note with the parsed TARGET_FLAGS.
I then use this note in my regression tests to check that the cflags have been correctly parsed. This approach didn't feel right, but I could see a better way.
Vasileios, I'm happy to add support for that option after this review.
Also added some support for quoted strings. I don't think my approach works on Windows shells in general, but I don't this that's of concern for LNT?
Thanks again for your review.
Lets do these imports at the top. I assume these are in a standard python 2.7 distro?