Ok, so let me give you more informations :

The values reported are BAD when fetched with the command (you can trust me on that) :

How can you explain the 'battery.charge_level.percentage' reported as '0' in the example with a 25% charged battery properly reported by the other command :

Since I also use the 'battery' applet v0.4 you patched to reflect values from the backported driver, I can afirm the values are FALSE ...

Easy to verify step by step, if I modprobe -r bq27x00_battery then on the desktop, after some seconds for the applet refresh, the value revert back to the bme value and report properly 25% of battery available, if I modprobe the backported driver then, the applet refresh and show 0% of battery available ...

I have installed the v47 since you released it, and no problem until today.

So the question is : What I have made today ?

Answer, here is how I triggered this BUG :

I have a desktop battery charger, and yesterday night I have omitted to charge my N900, so this morning I have powered off the N900 because I have no more battery power and the device reported 2%.

Then I have replaced the battery by the one fully charged by my desktop charger some days ago, and during today the problem occur ...

Since the problem occur, the value reported for parameter 'battery.charge_level.percentage' is '0' and this is the value used by the applet 'battery' you patched ...

How can you explain, today, my applet reported 0% battery available during all the afternoon ?

It look like bme values reported take care of this scenario (changing physically the used battery by one fully charged) where the new driver does not ...

So there is really a bug when we fetch information from the backported driver ...

A++