mosml as a front end for MLton

[Stephen Weeks]

> Perhaps I should study these hacks --- are they the ones Shao describes in 
> his paper?

I haven't read Shao's paper, so I don't know.  I developed MLton's mostly
independently, although I had read some paper by Macqueen on the SML/NJ 
implementation and there are some similarities with their implementation
(mostly in that there are "stamps" to do fast equality checking of signatures
and prevent copying them).

> BTW, why do you want to use the mosml front-end --- what advantages does it 
> offer you?

MLton doesn't have a decent front end right now.  There is almost no error
checking.  We believe it accepts a superset of SML, but it is not usable as a
standalone SML development environment.  We would like to make it so.  We were
considering using the Kit front end or the mosml front end, but neither seems
viable at the moment.