Timekeeping [Was: Re: cvs commit: src/usr.bin/vmstat vmstat.c src/usr.bin/w w.c]

[Marcel Moolenaar]

On Oct 21, 2005, at 12:12 PM, Poul-Henning Kamp wrote:

> In message <3F6E14D5-73B2-448A-9440-32DFFBF4E9C4 at xcllnt.net>,  
> Marcel Moolenaar
> writes:
>
>
>> Thus the question of "since when" can be answered as: the first
>> time it became operational after being non-operational.
>>
>
> Bad definition: it literally means it should start counting from
> the first boot of the operating system.

No I meant that uptime is defined in terms of the most recent
activation time, where activation is the switch from non-operational
to operational. The use of "first" was confusing as it would
imply there can be more, which there can't be. Bad wording...

> I think we need the definition to consider if (process- ?)state is
> retained while the system is unconcious or not.

I'm not sure. I think that might be what makes the definition
complex. The use of operational vs. non-operational in absolute
sense seems to avoid a lot of complexities without being
unnatural. Sure, the DDB case is weird, but we can all agree
that no new processes can be forked while in DDB (due to the
fact that no process is running). This, the running of processes,
is also part of the operational state of an operating system.
Even suspend and hibernation are covered quite naturally. We
only have to let go of boot or reboot as the beginning of the
operational state...

-- 
  Marcel Moolenaar         USPA: A-39004          marcel at xcllnt.net