Subject: Swap overcommit (was Re: Replacement for grep(1) (part 2))
To: Matthew Dillon <email@example.com>
From: Charles M. Hannum <firstname.lastname@example.org>
Date: 07/13/1999 19:05:29
> :That's wrong.
> :The application might be killed by SIGKILL on current FreeBSD
> :implementation, when the system becomes swap shortage.
> ... a situation which will never occur if you are managing the memory
> through your own custom library. Therefore not relevant.
That's a strawman, if not a complete non-sequitur. Having my
1000-hour simulation get the axe because someone accidentally(?) ran a
non-blessed program on the machine that doesn't happen to use the
`[my] own custom library' is simply unacceptable. And this is far
from a hypothetical situation.
Overcommit does not give applications the *choice* of dealing
gracefully, because they don't know they're screwed until it's too