Finalizers and FFI

Alastair Reid alastair at reid-consulting-uk.ltd.uk
Thu Jun 10 11:35:10 EDT 2004


> BUT: This can make some people unhappy. Isn't there a more deterministic
>   way to schedule finalizers?

I wrote the Hugs version which does try to be more deterministic so I'm 
probably not the best qualified to write about what's wrong with GHC :-)
But, I think part of the problem is that GHC gives the 'main' thread special 
status: the whole program quits when the main thread quits even if there are 
other threads that are runnable.  There'd be less of a problem if the whole 
program quit only when there were no runnable threads _and_ garbage 
collection did not create any runnable threads.

I think there's a reason why GHC gives the main thread special status though.  
They certainly went to some effort to give it special status so I guess there 
must be a good reason that they did so?

--
Alastair Reid


More information about the Glasgow-haskell-users mailing list