On 06/23/2010 08:15 AM, Dave Neary wrote:
Hi,
Rob Staudinger wrote:
On Wed, 2010-06-23 at 14:06 +0100, Dave Neary wrote:
Good point - it would be great to know when you're on battery, and do
even less.
What would be even more helpful than going back and forth
"apps should know about battery" ..
"no they should not" ..
"but they need to" ..
would be if people who have concerns about their apps would bring actual
use-cases forward.
I agree :) Didn't realise I was going to be starting a ruckus.
I agree with Arjen that if your app is going to be in a context where
power is important, then you should just make power usage a top priority
regardless of "on A/C" or "not on A/C". But with netbooks and mobile
phones, we'll clearly be in a situation where some portion of the usage
time will be on A/C and in those situations, saving battery is not an
issue (general "being a good citizen" still is, of course).
Once the tradeoff between user experience and power consumption is clear
for a certain case it will be much easier to decide whether AC/battery
status information is needed.
Here's a few examples that come to mind:
* Video player - wants to give best experience possible when on A/C, so
goes with HD, high frame rate - but when on battery, reduces video
quality& frame rate to use less energy
* Email application that checks for& downloads email every minute/5
minutes when you're hooked to A/C, but only every half an hour when on
the move
* Wifi detection that updates very regularly when on A/C, but only
occasionally (or after manual request) when on battery
* Screensaver that goes from "nice attractive eye candy" when on A/C to
"blank screen" (or suspend) when on the move
* Window manager automatically turning off knobs& whistles& 3d effects
when on battery
etc. etc.
There are lots of situations where when you're on A/C you expect instant
updates, and when you're on the move they're just not that important.
And those are opportunities to adapt your user experience to be less
featureful, and thus less power-hungry.
most of this list is not even applicable to handsets, and really, only
applicable to geekstations (let's just use that term for workstations
which never make it into C2 or deeper, and run some form of cows or seti).
Yeah, this post is a bit sarcastic, sorry about that :)
I can probably pick each item in this list apart easily. Wifi usually
updates ten times per second if not more, even with power management
features enabled. I don't think more updates per second will help anyone.
Even on my home geekstation with a 30 inch LCD screen, I want the damn
screen to go off so it doesn't suck 100W when I take a quick break :)
(Mind you, my wife has one too so it shows in the electricity bill if
the screensaver on one of them is not functional and lights up my living
room at night).
Window manager knobs and whistles would consume cpu power when doing
normal work, therefore degrading the performance of the system while on
A/C power, thus lowering the overall response and hindering other
applications - so window manager bells and whistles need to be fast and
power efficient (3d hardware accelerated, short, responsive ones offer
great return on that. blatant throwing cpu power at this is just going
to get in your way, and makes people use twm).
e-mail, now there is a valid point. The big problem is here that unix
mail technologies aren't very good at PUSH mail. Only IMAP supports it
and most clients ignore that. This is an issue with the underlying
system. But, then again, keeping a TCP connection open and once-a-minute
writing 30 characters to that is really not that power costly, is it?
almost always, in the long run, coding power-efficient algorithms as the
default and only solution, will work the best for everyone.
I have yet to see an algorithm where "if (on_ac_power) {" does something
anyone in my household excluding my 2 cats would really appreciate :)
BTW, the favorite spot in the house for one of my cats is on top of the
cabinet that holds our set-top-box for cable TV.... even when it's off,
it still produces a ton of heat.
I think the main problem with everyone coding for mobile is that they
see a "pot of gold" at the end of the rainbow when "coding for power
use" comes up, as if you can do amazing new things with a system when
you plug in (or unplug) the power cord. This of course is completely false.
Auke
_______________________________________________
MeeGo-dev mailing list
[email protected]
http://lists.meego.com/listinfo/meego-dev
