Yes, that's how the code is supposed to work.  I no longer remember
why the `npxproc != p' case is there at all; I note from the history
that the check was also there before I added the code to track the TS
bit, so it's probably just an artifact.  I admit I was lame about
updating some of the comments.