[PATCH] weston: Filter keys by id bit

[Martin Minarik]

> > The actual sequence:
> >
> > [2134420.1337]  -> wl_keyboard at 39.key(2316,
899782584, 30, 1)
> > [2134421.1337]  -> wl_keyboard at 39.key(2316,
899782584, 30, 1)
> > [2134422.1337]  -> wl_keyboard at 39.key(2316,
899782584, 30, 0)
> > [2134423.1337]  -> wl_keyboard at 39.key(2316,
899782584, 30, 0)
> >
> > The correct sequence - after applying this patch:
> >
> > [2134420.1337]  -> wl_keyboard at 39.key(2316,
899782584, 30, 1)
> > [2134423.1337]  -> wl_keyboard at 39.key(2316,
899782584, 30, 0)

> Sorry I completely cannot figure out any reason the
client can't do this
filtering.

it can't. because of the release 3. it is actually not the
last time that
key released. only compositor knows it's still held.

it's also the protocol I disagree with and needs to be
fixed too
 "physical state of a key which provoked the key event"
but it says "key is not pressed"
obviously the possibility of multiple devices was not
clearly explained here. the wl_keyboard was meant to
provide an abstraction of single device

please note that the compositor itself needs filtered keys
for the purpose of
constructing the keys stack, calculating the modifiers,
key bindings and for
grabs.

constructing the keys stack is necessary because it has to
deliver already
held keys on surface enter. It is constructing this stack
even before any client
comes to play.

the proposal is to simply extend this existing `filtering`
to delivered events.
mainly to clear those early releases. it's simply a side
effect it clears also
other stuff.

> And I suspect there are clients who would want to see
the multiple events.

if they are going to detect the multiple presses that way,
it will be completely
incorrect. they simply aren't supposed to do that. the
wl_keyboard is supposed
to look like single device from the client's side. if the
user needs to
distinguish multiple physical keyboards she needs:
 - assign them to multiple wl_seat
 - create a new api wl_keyboards(?), where key() involve a
counter how many times
   is key held. this counter will be a "replacement" for
enum key_state