time bug

[Matthew Fluet]

> > Steve, I'm also curious if you were able to recreate the bug on your
> > system with the mlton-20010321 version and the timeBug{1,2}.sml files.
> 
> Yes.  On my 400 MhZ machine, I see lots of messages printed for when GC > T.  On 
> my 733 MhZ machine, I consistently get a Time exception.

A Time exception?!  (I never got exceptions, just the messages.)  Do you
know what's raising it? Unless it's in some Time.from* function being
called for a conversion from a _ffi value, then it would appear to be
from
          val T = t' - t
          val GC = gc' - gc
but, regardless of how the underlying functions are implemented,
successive calls to the locally defined times() and time() should yield
monotonically increasing total time and total gc time.  My feeling was
that the bugs were related to the fact that the total time and the total
gc times weren't necessarily comparable.

> I also can't quite get them working with the modification of your code that I
> described yesterday.  I don't think it's worth figuring out, since we're going
> to switch to getrusage everywhere.  Matthew, if you haven't done so already,
> I'll take that on.

I haven't done anything more than think about what I sent out earlier
today.  So, feel free to take that on.