Byron wrote:
> you could also make runST core dump --- but you don't do that.
But that's not what the original runST paper suggested and it's
not at all useful whereas what's implemented does match the
original paper and is useful.
I'm not saying this is the only choice or even the best choice - just
that it's a reasonable choice and therefore not a bug.
> maybe someone can answer this.
>
> why cant runSt have the following type:
>
> runST All a . (Cxt1..Cxtn) => (All s . ST s a) -> a
>
> ?
Well, that's not a valid type - but it's clear what you mean.
It is possible to do this and (I just checked) Hugs 1.3c does so.
When we finish merging the new backend with the new typechecker
Hugs 1.4 will do so too. The Hugs news page gives some hints
about what else needs done before we'll have a releasable system.
Alastair
