Re: [gpm] Using HAL: reporting.low and reporting.warning

[Richard Hughes <hughsient gmail com>]

On Tue, 2006-01-10 at 20:54 +0100, Jaap Haitsma wrote:
> Richard Hughes wrote:
> > On Tue, 2006-01-10 at 04:38 +0100, Jaap Haitsma wrote:
> > 
> >>Richard Hughes wrote:
> >>
> >>>On Sat, 2006-01-07 at 05:11 +0100, Jaap Haitsma wrote:
> >>>
> >>>
> >>>>Hi Richard,
> >>>>
> >>>>My batteries have as properties in HAL a reporting.low and 
> >>>>reporting.warning (also charge_level.low and charge_level.warning and 
> >>>>these have the same number so I guess they are the same thing)
> >>>
> >>>
> >>>Yes, defined by ACPI, and exported by HAL. I'm not sure if these are set
> >>>at manufacture time, or change with the battery lifetime.
> >>
> >>These settings are fixed. Low or critical battery capacity remains low 
> >>no matter how old your batteries are.
> >>
> >>The thing that changes is the last_full capacity. These decrease due to 
> >>aging of the battery.
> > 
> > 
> > Hmm, it's a shame low and critical are not updated for the lifetime of
> > the battery.
> 
> Probably I wasn't clear
> 
> But they should not be updated. With for example 1000mWh left and a 
> latop taking about 10000mWh you'll have always 6 minutes of battery life 
> left. The thing that's going down is the full charge which might have 
> been 50000mWh when you buy them (so you 5 hours of battery time). The 
> full capacity might go down to 30000mWh after a year of heavy use which 
> gives you 3 hours of battery time. The critical threshold in that case 
> will still be 1000mWh which still gives you 6 minutes.

Ohh I see, thanks for the explanation.

> >>>>Can't we use these properties instead of the sliders? Or are these 
> >>>>properties not present for all type of batteries?
> >>>
> >>>
> >>>I think maybe these values could set the default for the sliders, but I
> >>>think the sliders should remain -- as people have different preferences
> >>
> >>Let me try to convince you that people do not have different preferences 
> >>here.
> >>
> >>When batteries are getting low people just want a warning which says 
> >>that their batteries are low and that they have let's say 10, 20 or 30 
> >>minutes more.
> > 
> > 
> > The thing is, I'm pretty good at ignoring the first few warnings, rather
> > than scrambling to find the ac_adapter. I'm sure other people are just
> > the opposite.
> > 
> > 
> >> If batteries are critical there should be a warning dialog 
> >>and the laptop should hibernate within 10 seconds if the user does not 
> >>cancel it. (We should still add this option that a user can cancel 
> >>hibernation, a user might want to hit the send button of his mail 
> >>program before hibernating )
> > 
> > 
> > We need to use the libnotify callbacks for this I think.
> > 
> > 
> >>The sliders with percentages do not work well if you for example now and 
> >>then use 2 batteries now or use batteries with different capacities.
> > 
> > 
> > Yes, this is a valid point. What about we just throw these sliders away,
> > and define a per-time warning, like a lot of people want.
> > 
> > As long as we have a sufficiently long warning time, compared to the
> > resolution of the updates and the changing of the cpu load, the aliasing
> > and inaccuracy shouldn't be a problem.
> > 
> > What about:
> > 
> > 1 notice at 20 minutes
> > 1 warning at 10 minutes
> > 1 critical warning at 5 minutes, with a link that lets the user abort
> > the action?
> > 2 minutes later, the action is performed regardless
> 
> I'm all for it with some minor changes
> 
> I'd say to make you notice things we should use modal dialogs when thing 
> s are getting critical.
> 
> So:
> 1 notification warning at 20 minutes
> 1 modal dialog at 10 minutes ( to make sure the user notices this)
> 1 critical warning at 5 minutes in modal dialog (will hibernate within 
> 10 seconds) with a button that lets the user abort the action
> 
> 2 minutes later a dialog pops up with a message that the system will 
> hibernate in 10 seconds. This gives the user a change to still do this 
> last minute thing.

Sure, I think we need to implement this only when the shutdown warning
are per-time, rather than per-percentage like we have at the moment.

> > Does this have to be configurable? 
> 
> No

Agreed, unless anyone has any objections?

Richard.