The Death of Finalizers

[George Russell]

Simon Marlow wrote:
> 
> > Do we still have atomicModifyIORef?
> >
> > I don't know that there's that much point now that Haskell finalizers
> > are gone.  I don't have a strong objection (esp. since it seems every
> > implementation now has it) but it doesn't seem so necessary now.
> 
> It ought to be significantly faster than using MVars if you can get away
> with it.  I haven't measured it though.
I bet it's a great deal faster than MVars on GHC if you also have to
take asynchronous exceptions into account.