SSA backend

[Stephen Weeks]

> Can you summarize the state of the SSA backend?  Is it faithful to the 
> description you sent out a couple of months ago, or has it evolved
> dramatically since then? 

It is much like what I described a couple of months ago - i.e. CPS without block
structure.  The only changes that I can think of are that overflowing primapps
are now treated as transfers (i.e. end of basic blocks) and that handlers are
explicitly stated at nontail calls, instead of inferred.

The code is in the CVS tree.  I'll tell you how to take a look once I get my CVS
mixup straightened out.