Cmm Commentary Update

[Peter Tanski]

> Peter Tanski wrote:
>> I updated the Cmm Commentary page, at last...

Simon Marlow wrote:
> Thanks Peter - very detailed.  I took a quick skim through and  
> clarified one small point.

Simon Peyton Jones wrote:
> That's great thank you. If it's getting big, split it up into sub- 
> pages.  I doubt I will look at it in the next few weeks, though --  
> I'm totally snowed.

It isn't as in-depth as I would like but it might help anyone new to  
Cmm.

SimonM, thanks for looking at it and making the clarification on  
inlining GlobalRegs over foreign calls--much better.

SimonPJ, I am moving the part on floating point exceptions from "Cmm  
Design: Observations and Areas for Potential Improvement" to a new  
page: "Cmm: Implementing Exception Handling."   There I will add some  
extra notes on implementing floating-point exceptions in Cmm, as this  
seems like an area that will require some design discussion.  (From  
what I can tell, a bit of this has already been set up at the Haskell  
level, through Control.Exception; interfacing with Control.Exception  
and design details particular to a machine would fill out the rest.)

Cheers,
Pete