[m-dev.] for review: add support for existential types [3/4]

[David Glen JEFFERY]

Fergus,

I'm happy with your responses to part 2 of my comments. Part 3 was generally
fine too, but I have a couple of things to add.

> > > +	% XXX we should check that `TVars' is of the
> > > +	%     appropriate form (i.e. a list of variables)
> > 
> > Does this get checked elsewhere?
> 
> No.
> 
> > If not, you should fix the code.
> 
> Yes, you are right <sigh>.

Can we see the diff for that bit?

> +		% (XXX is this check overly conservative? Perhaps we should
> +		% allow existential constraints so long as they contain
> +		% at least one type variable which is existentially quantified,
> +		% rather than requiring all variables in them to be
> +		% existentially quantified.)

We need to think about this some more. What did we decide the other day when
we talked about this?

So... can you please reply to my comments on part 4? When you have done so
and you have added the code (and shown the diff) for checking that the
TVars in a "some" are in the correct form , this should be ready to commit.


love and cuddles,
dgj
-- 
David Jeffery (dgj at cs.mu.oz.au) |  Marge: Did you just call everyone "chicken"?
PhD student,                    |  Homer: Noooo.  I swear on this Bible!
Department of Computer Science  |  Marge: That's not a Bible; that's a book of
University of Melbourne         |         carpet samples!
Australia                       |  Homer: Ooooh... Fuzzy.