Dr. David Kirkby wrote: > Hi, > I'll reply to several of your questions/notes in the one message. > > 1) I forgot to mention that if the laptop is booted without the battery, then > the GNOME battery charge monitor will never show anything useful at all. It > almost seems if the battery charge monitor decides the machine is mains > powered and does not bother checking any more. Booting without a battery also > removes the battery charge monitor applet from the task bar, but adding it > back does not help - it still indicates there is mains on the unit and no > battery present. > > So it seems a no-no to boot without a battery present. It would be good if > that could be fixed. >
Maybe this is all related. > 2) I understand Sony could use their own API's, but even if they do, I think > you might be able to solve it - see next point. > > 3) Your dtrace script generates output from both inserting AND REMOVING the > battery! So it seems dtrace is seeing things 'lshal -m' can not. I run both > 'lshal -m' and your dtrace script in separate terminals and found dtrace was > detecting both insertion and removal, but 'lshal -m' only shows the insertion > of the battery. > > Also, your dtrace script works whether or not the machine had a battery in > when it was booted. So it seems to get all relevant information under all > circumstances. > > # ./acpi.d > dtrace: script './acpi.d' matched 2 probes > > Battery is inserted, so I will remove it > > CPU FUNCTION > 1 -> acpi_drv_cbat_notify > acpica`AcpiEvNotifyDispatch+0x7d > 0x1 > 1 -> acpi_drv_cbat_notify > acpica`AcpiEvNotifyDispatch+0x7d > 0x80 > > Battery was removed and generated output above. Now I'll insert the battery > > 1 -> acpi_drv_cbat_notify > acpica`AcpiEvNotifyDispatch+0x7d > 0x0 > 1 -> acpi_drv_cbat_notify > acpica`AcpiEvNotifyDispatch+0x7d > 0x80 Good, there are events, but not consistent with what I have seen and not what the spec states. At least with any events, there could be a workaround. I'll send something to try later. > > The next time the battery is removed, I do not see 'CPU FUNCTION' again - I > assume this is some feature of dtrace - I've never used it before. > Yes, that header appears only once. > 4) I understanding predicting battery life can be difficult. I have 3 > batteries, so it's not as simple as calibrating the battery, unless the > serial number can be read from them and a log kept of the performance of each > battery. > > 5) If you need help with differential calculus, help might soon be at hand in > the form of Sage (http://www.sagemath.org/), which is a pretty impressive > sounding bit of maths software, which has: > > "Mission: Creating a viable free open source alternative to Magma, Maple, > Mathematica and Matlab." > > Sage has commercial sponsorship from both Google and Microsoft Research, but > is open-source and GPL'ed. Just what I want. A Solaris port is in progress > and might, with a bit of luck, be here this month. But it will be a few > months at least before I think it will be stable on Solaris, so don't rush > out to use it. > Nice. > 6) Does the current battery indicator switch the power off gracefully if the > battery is exhaused? The machine always used to power off ungracefully, with > no syncing of file systems. But I think the last time I used it, there may > have been a graceful shutdown, but I can't recall for sure. > There is no graceful shutdown, i.e. the battery monitor doesn't do anything. > 7) When I last checked the time for the battery monitor to register the > change of status, so it worked quickly a few times. Perhaps there was some > transient thing on the system which caused it to take a long time, but I'm > sure I've seen this issue before. > > It is is going to power the machine off gracefully, and there is a > significant problem with the monitor indicating the charge accuratly, it > might be better to have the facility to disable that, or let the user add > some time period after exhaution before the shutdown occurs. > > 8) When the machine is booted with a battery, and the applet knows this, it > indicates the charge OK. If the battery is then removed (which does not cause > any change on the aplet), but replaced by a *different* battery, the applet > clearly knows the battery is different and indicates the correct charge. I > have two batteries near me at the minute - one 100% charged, the other 86% > charged. The applet knows which is which, as long as the machine was booted > with a battery in place. > I'll get you something to try soon and hopefully the other issues may go away. Phi
