Comment 3 for bug 812394

Javier has asked me to comment on this. But my role here was only to specify how the policy should be presented in the power settings. John Lea proposed the policy and Mark Shuttleworth reviewed it.

As John has explained it to me, the logic behind the new policy is this:
1. In Ubuntu, Suspend and Hibernate often don't work on new hardware models without software fixes.
2. The nature of Suspend/Hibernate bugs is that while (as Martin points out) they can be fixed for individual machines, the only feasible way to prevent them from occurring, again and again with new hardware models, is for hardware to be certified.
3. For Ubuntu to present a Suspend function that doesn't work is unprofessional, and presenting a Hibernate function that doesn't work (and destroys data by never waking up) is even worse.
4. It is feasible for Ubuntu to tell whether the hardware it is running on has been certified.
5. Requiring users to understand that Suspend and Hibernate might not work before turning them on is also bad, but not as bad.
6. Suspend and Hibernate do work on many hardware models that have not been certified.
7. From (1), (2), (3), and (4), Suspend and Hibernate should not be available by default for hardware that is not certified.
8. From (5), (6), and (7), Ubuntu should let people turn Suspend and Hibernate back on again.

I can see various ways in which this policy could be shown to be unsound:
* Show that (1) is false. For example, Martin says that Suspend "works on the vast majority of machines". What percentage is that? Does anyone in the world know?
* Show that (2) is false, e.g. "XYZ new standard or technique means that Suspend and Hibernate will almost always work without certification".
* Show that (3) is false or partly false, e.g. "when Hibernate fails, N % of the time it fails before power off, so there's no data loss".
* Show that (4) is false, e.g. "millions of machines asking a server whether their hardware is certified wouldn't scale".