Holger Macht wrote:
> Hi developers,
>
> 1. Changing MODE_* to CAPABILITY_*
> ----------------------------------
>
> at the moment, we have five socalled "MODES" which clients use to connect
> to the daemon:
>
> /** Forward acpi event from /proc/acpi/events */
> #define MODE_ACPI_EVENTS 1
> /** Forward powersave events created by the daemon */
> #define MODE_POWERSAVE_EVENTS 2
> /** Forward notifications the daemon throuws */
> #define MODE_NOTIFICATIONS 4
> /** Forward screensaver requests */
> #define MODE_SCREENSAVER 8
> /** Forward progress bar requests */
> #define MODE_PROGRESS 16
>
> MODE_ACPI_EVENTS and MODE_POWERSAVE_EVENTS really does not have anything
> to do with a capability from clients. The daemon needs not to know whether
> some client uses acpi or powersave events. So at first I like to remove
> these two "MODES".
>
> Furthermore, I like to rename the remaining three, because MODE_* is not a
> descriptive define. Better is CAPABILITY_*, I think. I like to do it _now_
> because until now noone else (I hope) uses these defines.
>
> Also I like to rename MODE_SCREENSAVER to CAPABILITY_SCREENLOCK, because
> we do not need a nice screensaver. The only thing that matters is the
> locking of the screen.
>
> So these defines would look as follows:
>
> /** Forward notifications the daemon throuws */
> #define MODE_NOTIFICATIONS 1
> /** Forward screenlock requests */
> #define MODE_SCREENLOCK 2
> /** Forward progress bar requests */
> #define MODE_PROGRESS 4
>
>
> Additionally, I like to also adjust the client struct:
>
> typedef struct cNode {
> ...
> - /** The mode to which the message belongs */
> - int mode;
> + /** The capabilities to which the message belongs */
> + int capabilities;
> ...
> } cNode;
>
> The problem is that we have to adjust all clients to use the new
> defines(kpowersave, wmpowersave, testclient, dbus_testclient). But I also
> think that this is the last chance to do it.
>
>
> 2. Broadcast everything and always over the bus
> -----------------------------------------------
>
> Due to the fact we are using DBus now, we do not need a connected client
> to send out events, progress, whatever. So we can broadcast everything
> (all five MODES) over the bus. Clients still have to connect to take work
> off from powersaved. For instance taking care about screenlocking. But
> also without a connected client, all messages go over the bus.
>
>
> Please comment and say if I can do that. I've got the code already, so I
> can commit immediately.
>
This is your area, go ahead ...
Has anyone already measured how much CPU load these messages produce?
Could this get a problem?
Thomas
_______________________________________________
powersave-devel mailing list
[email protected]
http://forge.novell.com/mailman/listinfo/powersave-devel
