[Techtalk] Theory vs. practice

[Kai MacTane]

At 1/16/02 03:07 PM , Raven, corporate courtesan wrote:
>         Yike; it makes perfect sense, but I hadn't realized that.  So
>who watches the watcher, as it were?  I'm reasonably sure that it must
>be able to watch itself, because otherwise you'd have an infinite
>progression of watcher watchers, watcher watcher watchers, and so on.
>But how does it know that it's handling itself properly?

Simple, as long as each process can do more than one thing: each watcher 
watches two other processes (but not itself). Nobody needs to watch 
hirself, everybody gets watched, and there isn't one single process that 
has to watch everything and not get watched.

    (Sort of the way the checks and balances of the US federal
    government (theoretically) work. Each of the three branches
    (executive, legislative and judicial) watches the other two
    branches, and the people (again, this is theoretical) watch
    all three (and the executive branch, by design, *does* keep
    at least some tabs on the people).)

The final thing you need to make this system work is a device by which any 
new process is assigned its two watchers. Perhaps newly-spawned processes 
can pick a pair of watchers at random? Or maybe they just get auto-assigned 
the two most recent, or the two with the lowest load, or something.

                                                 --Kai MacTane
----------------------------------------------------------------------
"The seasons don't fear the reaper,
  Nor do the wind, the sun and the rain.
  We can be like they are."
                                                 --Blue Öyster Cult,
                                                  "Don't Fear the Reaper"