exceptions and coercions

[Stephen Weeks]

> I've got a question about exceptions though.  Given
> 
>   let x = e1 in handle z => e2
>   in  e
> 
> I have the flow rule for exceptions being:
> 
>   F(x) >= F(last(e1)) and F(last(e)) >= F(last(e2))

I have no idea why the second rule isn't

F(x) >= F(last(e2))

x is obviously a join point of e1 and e2.

> However, our notes indicate the closure converted
> translation as being:
> 
>   let y1 = [e1]
>   in X(y1,F(last(e1)),F(last(x)))  
> 
>   handle z =>
> 
>      let y2 = [e2]
>      in X(y2,F(last(e2)),F(last(x)))

This jibes with what I wrote above.