> Simon says:
> Warren makes an excellent point, though he doesn't highlight it:
>
> unlifted tuples are INCOMPATIBLE with seq
>
> The reason for this is that we can't implement unlifted tuples directly at
> all. Instead, we simulate them by making pattern matching lazy, so that
> _|_ behaves identically to (_|_,_|_), even though the two may have different
> runtime representations. The seq function exposes the trick!
In a private conversation with Arvind some days ago I pointed out
that, contrary to his earlier message, there is a situation where Id,
too, distinguishes lifted from unlifted tuples. It turns out to
correspond exactly to the situation that Warren and Simon cite. In Id
we have a sequencing construct (called a ``barrier'') which behaves in
the same way:
{ x = (e1,e2);
( y = x
--- % this sequences the statements in parens
z = y )
In
z }
This will return a tuple even if e1 and e2 diverge, whereas:
{ ( y = (e1,e2)
--- % this sequences the statements in parens
z = y )
In
z }
this will return a tuple only if e1 and e2 terminate.
Nikhil
