[coyotos-dev] Sleep, wakeup, and persistence

[Valerio Bellizzomi]

On 28/09/2007, at 20.39, Jonathan S. Shapiro wrote:

>On Thu, 2007-09-27 at 03:04 +0200, Valerio Bellizzomi wrote:
>> On 16/09/2007, at 14.44, Valerio Bellizzomi wrote:
>> 
>> >On 15/09/2007, at 19.33, Charles Landau wrote:
>> >
>> >>At 12:46 AM +0200 9/16/07, Valerio Bellizzomi wrote:
>> >>>The current sleep interface could have an additional parameter:
Enable
>> >>>wakeup on system restart,
>> >>
>> >>Use the Principle of Least Authority. If a program has no need to 
>> >>know about restarts, it should have a capability that has no option 
>> >>to reveal restarts.
>> >
>> >Ok. The sleep interface doesn't need the additional parameter, but the
>> >*schedule capability* should define if the program must be awaken on
>> >system restart.
>> 
>> More thinking on this question. It is a scheduling issue. There are the
>> process, the schedule capability, and the schedule which implements the
>> scheduling contract.
>> The schedule should obey what is defined in the schedule capability. If
>> the schedule capability has the option to reveal restarts, the process
>> should be allowed to react to restarts.
>> Note that no conclusions are drawn at the moment.
>
>I am unclear why you think this is a scheduling issue.

I think that the application's ability to react to restarts should be
defined in the scheduling contract.


Side note: I got a couple of commit mails, but the commits don't appear on
the web log, and make update finds nothing.


val