[Openmcl-devel] process-run-function and mach ports usage

[Ron Garret]

On Feb 24, 2011, at 8:36 AM, Tim Bradshaw wrote:

> On 24 Feb 2011, at 14:59, Raffael Cavallaro wrote:
> 
>> Hold on, do you mean to say that the wisdom of the market won't make the right decision here without some sort of regulatory intervention?
> 
> No.  The market considers it fine for a few people to die.  When lots of people start dying then things will get fixed, because that costs a lot of money.
> 
> (This is, partly, a joke).

Why?  This is exactly as it should be.  Just because people don't like to overtly face the fact that life has a finite value doesn't change the fact.

Also, let's not lose sight of the fact that this is OS X we're talking about here.  It's a consumer operating system.  It is not marketed for use in life-and-death situations.  It is not at all unreasonable for a consumer to accept having to reboot it every now and then in exchange for being able to acquire it for $100 instead of $1000.

It's actually far from clear that it is even *possible* to do much better in terms of reliability in a general purpose operating system.  My poster child is the Remote Agent Executive: it was mission critical, so significant resources were expended to make it reliable:  It was written in Lisp, it was formally analyzed and exhaustively tested and it *still* failed in flight.  So for you people lamenting the crappiness of the Mach kernel: what exactly do you propose we (the CS community) do differently?

rg