[Pm-utils] Removal of pm/power.d/sched-powersave

[Chase Douglas]

On Thu, 2010-05-27 at 20:39 +0100, Richard Hughes wrote:
> On 27 May 2010 20:30, Chase Douglas <chase.douglas at ubuntu.com> wrote:
> > I've been working on powersave policies in Ubuntu, where we've created a
> > package with a bunch of scripts. You can find some scripts I've been
> > working on for Maverick at:
> 
> I think a lot of the policy decision logic that depend on load or
> timings probably belong in the kernel.

Yes, the kernel probably could do more to handle adaptive policies, but
I'm not sure there would be consensus on how to do that across
distributions and use cases. I'm of the mind set that the kernel should
be a low level implementor, and userspace should tell it exactly what it
wants.

That said, I do completely understand that there are holes in the
scheduling powersave script I've been working on. If the load changes
after the power is plugged or unplugged, the policy will no longer be
correct. However, it's one degree better to check the current load
before doing any scheduling stuff.

> I also think as a general note
> it makes a lot of sense to keep things in pm-utils generally, as the
> hal and hal-info saga taught me it's very hard to maintain
> compatibility between policy and data, and getting things upstream is
> a sure way of "doing things properly".

I can understand both points, but what if we had two separate packages
upstream, one for policy and one for implementation (as pm-utils is for
the most part).

I'm certainly not opposed to working on upstreaming policies that others
find useful, but I also don't want to be forced into hacking around
policies that came as part of pm-utils itself.

-- Chase