bug: mlton: simplifyPrimExp: Case2

[Stephen Weeks]

>   I encountered a bug in MLton 20000906, and pruned the problem down
>   to the following code:
> 
>     % cat mltonbug.sml
>     val L = case Word32.wordSize 
>               of 32 => ["CAFEBABE", "FFFFFFFF", "A123"] 
>                | _ => []
> 
>   This looks silly on its own, but you'll have to believe that it came
>   from a sensible unit test.  :)

Thanks for the bug report.  I observe the same problem.  It was definitely fixed
by 20010125.  I guess the fix just didn't make the CHANGES file -- maybe because
we'd never noticed the bug.

>   Incidentally, this came from a little project (JVMkit, for
>   processing Java class files) where I'm trying hard to support all
>   the different SML systems (NJ, Moscow, Poly, MLton).  

Definitely a challenge.  You might take a look at how the Isabelle people
(http://www.cl.cam.ac.uk/Research/HVG/Isabelle/index.html) do it, since they
support all of the systems except for MLton (because it doesn't have "use").  I
think they have per system code to make the "standard" basis look the same in
all systems.

Let us know if you have any other problems, with 20000906 or 20010125.  We'll
have a new official release RSN :-)