Comment 121 for bug 629258

Brian: I am using the patched upower and friends from your PPA, and I have a regression: the estimated time is frequently calculated incorrectly, due to an incorrect calculation of the current. For example, in typing this comment, I watched as three notifications went by, all saying "laptop battery low" and reporting the battery at 5%, but in order, they reported estimated remaining times at 0:12, 0:06, and now 0:20. My power usage has not changed significantly during this period. I believe this is due to a sampling error, though I haven't looked at the patch so I can't say how often it really occurs or exactly how to fix it. In most cases it is fine though, so thank you for the work, but I think it needs further review and testing before it is merged into mainline. Furthermore, Intel's powertop gets the current correctly 100% of the time, maybe it is worth looking in to how they calculate it.

Also, I can say for certain that my device used to properly report current_now, in fact, I had scripts that used this value before and they worked perfectly. I wish I could say with certainty when this regression occurred, I expect it was due to a kernel update. I know it worked some time after I upgraded to maverick, but before I started using the 2.6.36 kernel from kernel-mainline (because that was supposed to fix the extra load balancing ticks, which it didn't really do, sadly). I don't expect that this is the issue though, since so many others see this bug that are probably not using the 36 kernel. I hope to do more investigation in the future.