Rate guessing code

[Richard Hughes]

On Fri, 2005-11-04 at 09:29 +0100, Søren Hansen wrote:
> fre, 04 11 2005 kl. 00:47 -0500, skrev David Zeuthen:
> > Now, I wonder if using even a higher-order method would be useful?
> 
> I thought about that. I'm however reluctant to take more than the two
> previous measurements into account. Traditionally the remaining time
> estimate says how long you'll be able to run with the current load, not
> the average load over the last x minutes.

For batteries with broken rate handing (i.e. that produce a rate value
that is wildly inaccurate) your method to determine the time remaining
would be much more accurate. Can we somehow test your fallback against
the reported rate (and use yours automatically if the rate differs by
too much) or is this too much overhead?

Richard.