Merge aura::DispatcherLinux into base::MessagePumpAuraX11.

Merge aura::DispatcherLinux into base::MessagePumpAuraX11.

The majority of incoming XEvents are related to specific XWindows and need to
be sent to a specific MessagePumpDispatcher. MessagePumpAuraX11 would pass the
XEvent to aura::DispatcherLinux, which would pass it to the next
dispatcher. Because of the chromeos unit tests, which spin up a new XDisplay on
each test, these two objects need to have equivalent lifetimes. Due to the
incoming clipboard support (which this patch was split off from),
DispatcherLinux can't live in aura:: anymore. The solution is to merge the two
objects that have to have equivalent lifetimes together.

The one ui:: specific check was split off into an MessagePumpObserver. This also got rid of those hacks where we would static downcast to DispatcherLinux
from MessagePumpDispatcher.

BUG=130805
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=154174

[sadrul, 2012-08-29 16:01:25]

I have left a few comments. LGTM

http://codereview.chromium.org/10895020/diff/7005/base/message_pump_aurax11.cc
File base/message_pump_aurax11.cc (right):

http://codereview.chromium.org/10895020/diff/7005/base/message_pump_aurax11.c...
base/message_pump_aurax11.cc:49: // MessagePumpAuraX11::Current()->display()
once GTK goes away.
Perhaps say once 'MessagePumpGtk' goes away (I was almost going to point out
that GTK is already gone from chromeos) :)

http://codereview.chromium.org/10895020/diff/7005/base/message_pump_aurax11.h
File base/message_pump_aurax11.h (right):

http://codereview.chromium.org/10895020/diff/7005/base/message_pump_aurax11.h...
base/message_pump_aurax11.h:50: unsigned long x_window);
Use XID?

http://codereview.chromium.org/10895020/diff/7005/base/message_pump_aurax11.h...
base/message_pump_aurax11.h:56: void
RemoveDispatcherForRootWindow(MessagePumpDispatcher* dispatcher);
It makes sense to explicitly state in the comments that there can be multiple
dispatchers for the root-window, but a single dispatcher for a normal window.

http://codereview.chromium.org/10895020/diff/7005/ui/aura/device_list_updater...
File ui/aura/device_list_updater_aurax11.cc (right):

http://codereview.chromium.org/10895020/diff/7005/ui/aura/device_list_updater...
ui/aura/device_list_updater_aurax11.cc:24: return base::EVENT_HANDLED;
I am not sure if returning _HANDLED here is OK. There are other observers (e.g.
XInputHierarchyChangedEventListener) that listens to this event, and we probably
want to keep sending the event to them.

http://codereview.chromium.org/10895020/diff/7005/ui/aura/env.cc
File ui/aura/env.cc (right):

http://codereview.chromium.org/10895020/diff/7005/ui/aura/env.cc#newcode123
ui/aura/env.cc:123: base::MessagePumpAuraX11::Current()->AddObserver(
Perhaps add a note here that adding the device-list-updater as a dispatcher for
the root-window doesn't work because the XI_HierarchyChanged event does not
include a target window.

[Elliot Glaysher, 2012-08-29 16:50:07]

sky: aura stamp
willchan: base stamp

[sky, 2012-08-29 20:50:34]

LGTM

[willchan no longer on Chromium, 2012-08-29 23:35:57]

base/ lgtm

[commit-bot: I haz the power, 2012-08-29 23:37:33]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/erg@chromium.org/10895020/9006

[commit-bot: I haz the power, 2012-08-30 03:03:40]

Try job failure for 10895020-9006 (retry) on win_rel for step "browser_tests".
It's a second try, previously, step "browser_tests" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...