Re: [Utopia] Re: [Gnome-Power-Devel] power mgmt in GNOME; requirements?

[David Zeuthen <david fubar dk>]

On Wed, 2005-04-20 at 00:01 +0300, Paul Ionescu wrote:
> On Tue, 19 Apr 2005 15:52:51 -0400, David Zeuthen wrote:
> 
> > [1] : g-p-m should simply choose itself what sleep state to put the
> > computer in; over time, for ACPI systems this may be S1 (sleep) -> S3
> > (deeper sleep) -> S4 (suspend to disk AKA hibernate) after SLEEP_TIME,
> > 2*SLEEP_TIME and 3*SLEEP_TIME.
> > 
> > Either way, the point is that the user will have difficulty figuring the
> > difference between "Suspend" and "Hibernate" and frankly he should never
> > have to spend time worrying about the difference.
> >
>  
> I disagree here. First a lot of users know the difference between
> sleep/suspend and hibernate because of M$ Windows which already has this
> difference, and second, there are cases when this is not what a user
> wants. There are times when I want only suspend (S3), and times when I
> want only hibernate. For S1 I don't care too much.

If STD and STR works well people shouldn't need to care about the
differences. Also, I disagree that people grasp this just because
Windows uses these terms. I can produce several Windows users who don't
and I'm sure that you can do the same :-) Also, do you really want
people to care about these differences?

We need, as much as we can, to do our design based on the ideal system
not around the stupid bugs or deficiencies we're seeing right now. If we
do the latter we've already lost.

> > [2] : ideally I wouldn't have them but just "do the right thing" when the
> > lid is closed (suspend etc.) but unfortunately we cannot assume that since
> > distros want to ship g-p-m with sleep completely disabled since ACPI on
> > Linux blows out of the box.
> 
> Some people close the lid because they use external monitors, so is bad to
>  assume that if you close the lid, the g-p-m should suspend the system.
> And then SuSE has ACPI S3 and S4 enabled for some time, and most
> others have at least S3.

We should just detect when an external monitor is plugged in. That won't
happen before the stuff Jon Smirl is working on lands in the Linux
kernel though.

David