haskell finalisers driven underground

[Sven Panne]

[ Best read while listening to Queen's "I'm going slightly mad" ]

Before this futile discussion starts over again, everybody should
really read Boehm's "Destructors, Finalizers, and Synchronization"
(as suggest by SPJ IIRC):

   http://www.hpl.hp.com/techreports/2002/HPL-2002-335.html

Even with real concurrency, finalizers are a nightmare to implement
correctly. (If you think differently and have some JVM background,
please send your resume to the company I'm working for. :-}

Cheers,
   S.