On 10/16/07, Linus Torvalds <torvalds@linux-foundation.org> wrote:
quoted text
>
> On Tue, 16 Oct 2007, Dmitry Torokhov wrote:
> >
> > I agree that these are 2 different events. My argument is that
> > "VOLUME_UP_NOTIFY" event is similar to "BATTERY_OUT_NOTIFY",
> > "DOCK_UNDOCK_NOTIFY", etc, etc and should be sent not through input
> > layer but through a generic (yet to be designed) notification
> > mechanism. Something lighter than input. Something like uevents over
> > netlink.
>
> Well, I'd argue that:
>
>  - it's going to be the same entity that cares in both cases (ie anybody
>   who is ready to accept VOLUME_UP keypresses is also the exact same
>   party that also wants to know if VOLUME_UP happened *independently*)
>
>   Ergo: making it a separate "generic" notification is actually totally
>   counterproductive, because it just adds complexity.

That is a good argument. If there are no other users for this other
transport then I agree, inventing it just for keypress notifications
is bad idea.

quoted text
>  - it really is a keypress. Yes, it's a keypress with side effects, but
>   it still tends to have a distinct source, and as such it is interesting
>   *as* a keypress.
>
>   IOW: it should have all the same "incidental" side effects as any other
>   keypress. Example: I think it's reasonable to consider it an event as
>   far as the screen saver is concerned. In other words, it's not *just* a
>   "volume was raised" event. It's a "volume was raised, and the user
>   actually pressed a key to do so".
>

This is a good argument for having 2 separate types of events but not
for which transport shoudl be used to deliver it.

quoted text
> So I do think they are keypresses, although I also suspect that like many
> other magical keys, the "NOTIFY" version is often also totally hidden by
> hardware/firmware interactions (ie I'm pretty sure that many of those
> special keys will never be visible at all to the OS, because the firmware
> hides the fact that they were pressed entirely!)
>

Yes, on my old Inspiron brightness is completely handled by firmware.
There is no ACPI, nor keyboard events generated whatsoever.

OK, I guess I would like to hear once again from userspace guys -
DBUS/HAL/etc. Do they see a need for a generic interface that can be
used for all kinds of events, not only related to keypresses. If they
say that they only care about notifications arising from keypresses
then I will add EV_NOTIFY type of events to input layer. What events
would we need? I can imagine:

KEY_BRIGHTNESSUP_NOTIFY
KEY_BRIGHTNESSDOWN_NOTIFY
KEY_BRIGHTNESSMIN_NOTIFY
KEY_BRIGHTNESSMAX_NOTIFY
KEY_VOLUMEUP_NOTIFY
KEY_VOLUMEDOWN_NOTIFY
KEY_MUTE_NOTIFY
KEY_DISPLAYSWITCH_NOTIFY
KEY_WIFI_NOTIFY

What else? And it better have "key" in its name, BATTERY_OUT_NOTIFY
won't fly ;)

-- 
Dmitry
-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/