Removes aura dependencies from WindowPositioner

Removes aura dependencies from WindowPositioner

This introduces WmGlobals. I would have liked to not have a static
Get(), but it's necessitated by some places not having
context. Hopefully I can change that later.

I'm not moving WindowPositioner to common yet as it uses
window_state. Once I move that I can move WindowPositioner.

For the time being I'm keeping the tests in ash as they need aura and
I don't want ash/wm/common depending upon aura.

BUG=603369
TEST=none
R=oshima@chromium.org

Committed: https://crrev.com/984c189ec33e36d863147f278e8f4db10cae691a
Cr-Commit-Position: refs/heads/master@{#388366}

[oshima, 2016-04-18 18:09:18]

lgtm

If we can live without Get(), that's good, but let's not introduce another
Shell, which increases dependency.

https://codereview.chromium.org/1900443002/diff/100001/ash/wm/aura/wm_window_...
File ash/wm/aura/wm_window_aura.cc (right):

https://codereview.chromium.org/1900443002/diff/100001/ash/wm/aura/wm_window_...
ash/wm/aura/wm_window_aura.cc:206: wm_windows[i] = Get(aura_windows[i]);
Looks like we're converting aura window -> wm window in several places. It may
be useful to have a utility function to convert from  vector<aura::Window> to
vector<WmWindow>

[sky, 2016-04-19 00:02:12]

I had two bugs in this refactor:
1. gfx::Screen* argument wasn't being used in WindowPositioner. I wasn't using
the screen arg as I want to route through the WmWindow impl which will have the
context to know the screen. To fix it I removed the screen arg and made the test
set the screen instance it was creating.

2. I need to call GetWorkAreaForWindowInParent in WindowPositioner, which has
special case code for windows.

Diff between 6 and 8 to see the changes.

Take another look?

[sky, 2016-04-19 13:11:32]

On 2016/04/19 00:02:12, sky wrote:
> I had two bugs in this refactor:
> 1. gfx::Screen* argument wasn't being used in WindowPositioner. I wasn't using
> the screen arg as I want to route through the WmWindow impl which will have
the
> context to know the screen. To fix it I removed the screen arg and made the
test
> set the screen instance it was creating.
> 
> 2. I need to call GetWorkAreaForWindowInParent in WindowPositioner, which has
> special case code for windows.
> 
> Diff between 6 and 8 to see the changes.
> 
> Take another look?

Diff between 6 and latest to see changes.

[oshima, 2016-04-19 18:11:52]

https://codereview.chromium.org/1900443002/diff/100001/ash/wm/window_position...
File ash/wm/window_positioner.cc (right):

https://codereview.chromium.org/1900443002/diff/100001/ash/wm/window_position...
ash/wm/window_positioner.cc:72: // On Win 8, the host window can't be resized,
so use window's bounds instead.
Are we still running win_ash in metro mode? Will running on desktop mode fix
this?

[sky, 2016-04-19 18:15:14]

https://codereview.chromium.org/1900443002/diff/100001/ash/wm/window_position...
File ash/wm/window_positioner.cc (right):

https://codereview.chromium.org/1900443002/diff/100001/ash/wm/window_position...
ash/wm/window_positioner.cc:72: // On Win 8, the host window can't be resized,
so use window's bounds instead.
On 2016/04/19 18:11:51, oshima wrote:
> Are we still running win_ash in metro mode? Will running on desktop mode fix
> this?

I think we still want this code for windows. Otherwise we use the actual work
area rather than from the window showing the ash environment.

[oshima, 2016-04-19 19:29:15]

On 2016/04/19 18:15:14, sky wrote:
>
https://codereview.chromium.org/1900443002/diff/100001/ash/wm/window_position...
> File ash/wm/window_positioner.cc (right):
> 
>
https://codereview.chromium.org/1900443002/diff/100001/ash/wm/window_position...
> ash/wm/window_positioner.cc:72: // On Win 8, the host window can't be resized,
> so use window's bounds instead.
> On 2016/04/19 18:11:51, oshima wrote:
> > Are we still running win_ash in metro mode? Will running on desktop mode fix
> > this?
> 
> I think we still want this code for windows. Otherwise we use the actual work
> area rather than from the window showing the ash environment.

The problem is (was) that metro mode always creates full screen even if
ash/chrome specifies the size, say 800x600,
which causes inconsistency between the actual window size and what we set to the
display.

Ash should be using the ash's display not native one even on windows. Isn't that
the case now?

[sky, 2016-04-19 22:13:14]

Patchset #9 (id:160001) has been deleted

[sky, 2016-04-19 22:17:22]

On 2016/04/19 19:29:15, oshima wrote:
> On 2016/04/19 18:15:14, sky wrote:
> >
>
https://codereview.chromium.org/1900443002/diff/100001/ash/wm/window_position...
> > File ash/wm/window_positioner.cc (right):
> > 
> >
>
https://codereview.chromium.org/1900443002/diff/100001/ash/wm/window_position...
> > ash/wm/window_positioner.cc:72: // On Win 8, the host window can't be
resized,
> > so use window's bounds instead.
> > On 2016/04/19 18:11:51, oshima wrote:
> > > Are we still running win_ash in metro mode? Will running on desktop mode
fix
> > > this?
> > 
> > I think we still want this code for windows. Otherwise we use the actual
work
> > area rather than from the window showing the ash environment.
> 
> The problem is (was) that metro mode always creates full screen even if
> ash/chrome specifies the size, say 800x600,
> which causes inconsistency between the actual window size and what we set to
the
> display.
> 
> Ash should be using the ash's display not native one even on windows. Isn't
that
> the case now?

Good point. Fixed.

Turns out the reason windows was failing a couple of tests was because the
window was getting sized differently, causing tests that expect certain sizes to
fail. I had to fix this to get the tests to work now (see win_window.cc).

Take another look?

[oshima, 2016-04-19 23:54:57]

lgtm

[sky, 2016-04-19 23:55:17]

The CQ bit was checked by sky@chromium.org

[commit-bot: I haz the power, 2016-04-19 23:55:48]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1900443002/180001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1900443002/180001

[commit-bot: I haz the power, 2016-04-20 00:00:47]

Committed patchset #9 (id:180001)

[commit-bot: I haz the power, 2016-04-22 19:18:27]

Description was changed from

==========
Removes aura dependencies from WindowPositioner

This introduces WmGlobals. I would have liked to not have a static
Get(), but it's necessitated by some places not having
context. Hopefully I can change that later.

I'm not moving WindowPositioner to common yet as it uses
window_state. Once I move that I can move WindowPositioner.

For the time being I'm keeping the tests in ash as they need aura and
I don't want ash/wm/common depending upon aura.

BUG=603369
TEST=none
R=oshima@chromium.org
==========

to

==========
Removes aura dependencies from WindowPositioner

This introduces WmGlobals. I would have liked to not have a static
Get(), but it's necessitated by some places not having
context. Hopefully I can change that later.

I'm not moving WindowPositioner to common yet as it uses
window_state. Once I move that I can move WindowPositioner.

For the time being I'm keeping the tests in ash as they need aura and
I don't want ash/wm/common depending upon aura.

BUG=603369
TEST=none
R=oshima@chromium.org

Committed: https://crrev.com/984c189ec33e36d863147f278e8f4db10cae691a
Cr-Commit-Position: refs/heads/master@{#388366}
==========

[commit-bot: I haz the power, 2016-04-22 19:18:28]

Patchset 9 (id:??) landed as
https://crrev.com/984c189ec33e36d863147f278e8f4db10cae691a
Cr-Commit-Position: refs/heads/master@{#388366}