Wow, there's already a thread for my app. =)

I'll spend my reply on answering some of the questions on this thread.

(1) What is being measured, and how?

As it has already been pointed out, the data is collected with lshal. The daemon listens for the OS to report that the values have changed (there's a DBUS signal when HAL updates the values), and records the new values and current time. So the raw data it collects is a group of points. The graph in the GUI is created by simply connecting the points with straight lines.

Now, everyone who knows something about scientific data visualization also knows that simply connecting the dots is just plain wrong, as it makes the visualization to suggest that we would know something about what has been happening in between the observations. I know that too, but it just makes for a nicer graph. =) And for the most of the time, if the data collection is continuously running, the actual values are not that much off, and the connected line gives an ok estimation.

However, when the data collection is not active (e.g. phone off), connecting the dots from that time might suggest that we can also record accurate data from a power down period. Obviously, there is no data being recorded when the device is not powered, it just looks like that in the graph. In a future version of the app (likely the next one) this effect will be fixed by not drawing the graph for periods when the daemon has not been running.

As you might expect, a 3rd party application like this shouldn't be relied to give 100% accurate measurements. So don't depend your (battery) life solely on this app.

If someone cares about the raw data, it is being stored in an sqlite db in /home/user/.config/battery-eye/ . If you know your SQL, the schema should be simple enough to work with.

(2) Does it shorten my battery life to I use the app?

In short, yes. Everything you run on a battery powered device will consume power from the battery.

However, the background data collection is designed to use minimal amount of resources (it only wakes up on DBUS signals and an infrequent timer event (60min) and does nothing else than record the current battery levels). In practise, I haven't noticed any difference in my battery life.

Naturally, if you use the app's GUI a lot (I like scrolling the graph back and forth a lot ;-) ), that will naturally have an effect.

(3) My battery doesn't charge to 100%?

Don't blame me, the program just reads the values from HAL as they are. If HAL says 97 percent, then that's what gets recorded and eventually displayed.

(4) My graphs look strange?

Yes, sometimes they do, especially the voltage graph. It seems that the OS updates the battery values more frequently when the device is actively used, and not at all when the device is completely idle. The voltage graph illustrates this effect quite well (first a straight 6~8hr line when I'm sleeping, and then a tight zigzag when I start to read my e-mails and slasdot in the morning).

At some point, I might try to figure out if there's a way to force the OS to update the values in HAL, without consuming too much battery. That might improve graphing accuracy from long idle periods.

(5) It seems that there are features missing?

Yes, it's still under development. I myself don't consider it ready.

BTW, the usual pre-extras disclaimer apply for battery-eye:

It's still in extras-testing, it may screw up your device!