> it seems to me that it'd be way easier to find all the races if we had
> a way to do a non-parallel make with the parallelizable targets in reverse
> order from how they are listed in the Makefile.  that should catch the
> vast majority of races.  I imagine there's some way to do this in the .mk
> files, but I'm not fluent enough in make to know how.  anyone know how to
> do this?

Last time I fiddled with make(1), I hacked on the list libs to have
a `Lst_DeQueue_Last()' and used that with the `toBeMade' list.

That caught a number of races like this.  However, I gave up when it
tripped on sushi...

-pk