time bug

[Stephen Weeks]

> I got that time bug again.  I've given up on stracing self-compiles
> (the compiler get < 50% of the CPU time and it's just not worth it).
> Anyways, here's the Failure message:
> 
> mlton: traceBatch: t' - t::-: t1 = 597.600 t2 = 598.590
> 
> And here's all the relevant code; I don't see anything wrong -- anyone
> else see something?

I read through it all and don't see anything.  I have a couple of suggestions.

1. Since Time.+ raises Overflow, not Time, your handle Time will do no good
   in that case.

2. In your tests, try printing out the arguments and results of Time.+ as used
   within the Control.time() function.