[gclist] Timeliness of finalization

Marc Shapiro shapiro@prof.inria.fr
Tue, 1 Apr 1997 14:37:34 +0200

 || From: stuart <stuart@cosc.canterbury.ac.nz> (yeates)
 || Date: Sat, 29 Mar 1997 14:27:39 +1200
 || Subject: Re: [gclist] Timeliness of finalization
 || Consider the case of a power-manager, which senses has just been notified
 || of imminent failure of the power supply, This situation usually leads to
 || the OS shutting down or crashing. If we are relying on finalisation to
 || ensure closure of files/consistency of disks, to maintain contracts in a
 || distributed environment or similar we still need to have our finalisers
 || invoked if at all possible.

Now that's a silly way to design an OS.  You need to flush your buffers to
disk FAST, i.e. in bounded time, where the bound is smaller than the latency
of your power supply.  No matter how timely finalization is, you can't know in
advance how many finalizers will be queued when the "power low" interrrupt
hits you, so you can't bound the latency.