[10:31] <xnox> mpt: it looks like we have reliable firmware hibernate on intel now. http://summit.ubuntu.com/uds-1308/meeting/21883/foundations-irst-support/
[10:32] <xnox> mpt: and it seems like it's what apple has been doing for a while now.
[10:32] <xnox> mpt: i'd love to discuss with you settings & installer changes that may be needed to accomodate for this. Ideally, I wouldn't want to introduce any UI, but we might have to.
[10:33] <mpt> ok
[10:34] <xnox> in essence - laptop goes to sleep, then there is a timeout (can be customized from 0 to a few hours, intel recommends default of 1h), after which everything from RAM is dumped onto a special partition and computer hibernates fully with like no power used.
[10:35] <xnox> mpt: on wake-up, firmware loads stuff from that partition into RAM and Ubuntu thinks it "got woken up from suspend". When actually it's been hibernating for days =)))
[10:35] <xnox> mpt: i think it will only take 30m - 1h to discuss, do you have time to schedule that sometime?
[10:36] <xnox> there are some privacy & encryption concerns but otherwise it's fairly straight forward.
[10:43] <mpt> xnox, I think I might be able to slot in some time to do that in November 2007: https://wiki.ubuntu.com/ExitStrategy?action=AttachFile&do=view&target=power-saving.jpg
[10:45] <xnox> mpt: excellent. "When asleep, go into zero-power mode ("hibernate")" can have "Never" as well right?
[10:45] <mpt> yes
[10:45] <xnox> awesome.
[10:45] <mpt> xnox, needs a bit of tweaking to fit into the current design, though :-)
[10:46] <xnox> mpt: and i think we may need to force disable it somehow. So for example - if one is using full disk encryption, asleep will have encryption keys stored in RAM then we must prevent it from going into hibernate, cause then encryption keys will be written out in plain text on the harddrive.
[10:47] <xnox> .... unless one has a hardware self-encrypting drive or ATA disk password.
[10:48] <xnox> and this technology is only available with SSD & <<8GB of RAM & when partition is present, thus for the rest we will continue to not offer "hibernate".
[11:09] <xnox> on the installer side - we will try hard to creae 8GB partition, when we believe intel rapid start is supported, but we might fail. In that case, we will offer manual force preseed to enable/disable creating that partition. But should that be exposed to the user?
[11:14] <xnox> mpt: how does this https://wiki.ubuntu.com/Power#PC relate to 2007 design?
[11:14] <mpt> xnox, it does not. ExitStrategy was a grand simplification that never happened.
[11:17] <xnox> mpt: fast user switching - done. Suspend & hibernate combined, will be possible with intel rapid start technology (that's what it is). And for the rest we don't expose hibernate (unless they follow my askubuntu and/or documentation to force enable it, in which case they keep both pieces). And the combined sleep/hibernate settings will be needed for the rapid start.
[11:18] <xnox> i'm not convinced about shutdown after in-activity, even desktops should sleep & transition to hibernate (there is desktop support for rapid start).
[11:25] <mpt> xnox, sleep/hibernate instead would certainly solve the problem of what to do with unsaved data.