[PATCH 2/11] ozone: evdev: Move KeyboardEvdev usage during dispatch to EventFactoryEvdev

ozone: evdev: Move KeyboardEvdev usage during dispatch to EventFactoryEvdev

This is part of a series moving IO on evdev devices to a fast thread,
since UI thread is too slow for the cursor.

The keyboard layout is a UI thread object, so move all of the keyboard
processing that needs layout information outside of the device-specific
processing. Devices no longer get a KeyboardEvdev* since they will
execute on a different thread.

BUG=449710
TEST=boot link_freon & press key

Committed: https://crrev.com/7b5b72bda2a843c48cb268181382d65f668c78ba
Cr-Commit-Position: refs/heads/master@{#313397}

[spang, 2015-01-23 21:56:57]

spang@chromium.org changed reviewers:
+ alexst@chromium.org

[alexst (slow to review), 2015-01-26 22:18:25]

https://codereview.chromium.org/872883003/diff/20001/ui/events/ozone/evdev/ev...
File ui/events/ozone/evdev/event_dispatch_callback.h (right):

https://codereview.chromium.org/872883003/diff/20001/ui/events/ozone/evdev/ev...
ui/events/ozone/evdev/event_dispatch_callback.h:29: typedef
base::Callback<void(int device_id, unsigned int code, bool down)>
My brain wants more structure here. 

You are passing things individually in one place, but as a struct in another. I
know there is an interface down the road that replaces these, but similarly, it
splits between the two styles.

I gather it's a callback param limit thing, but since we are bumping against it
in one place, would you hate me if I asked we moved all of these to structs?

[spang, 2015-01-26 23:01:22]

https://codereview.chromium.org/872883003/diff/20001/ui/events/ozone/evdev/ev...
File ui/events/ozone/evdev/event_dispatch_callback.h (right):

https://codereview.chromium.org/872883003/diff/20001/ui/events/ozone/evdev/ev...
ui/events/ozone/evdev/event_dispatch_callback.h:29: typedef
base::Callback<void(int device_id, unsigned int code, bool down)>
On 2015/01/26 22:18:25, alexst wrote:
> My brain wants more structure here. 
> 
> You are passing things individually in one place, but as a struct in another.
I
> know there is an interface down the road that replaces these, but similarly,
it
> splits between the two styles.
> 
> I gather it's a callback param limit thing, but since we are bumping against
it
> in one place, would you hate me if I asked we moved all of these to structs?

We could remove the structs once the 7 argument limit is removed. There is a
patch for this in review.

Why make more structs just to remove them?

[alexst (slow to review), 2015-01-27 13:28:24]

On 2015/01/26 23:01:22, spang wrote:
>
https://codereview.chromium.org/872883003/diff/20001/ui/events/ozone/evdev/ev...
> File ui/events/ozone/evdev/event_dispatch_callback.h (right):
> 
>
https://codereview.chromium.org/872883003/diff/20001/ui/events/ozone/evdev/ev...
> ui/events/ozone/evdev/event_dispatch_callback.h:29: typedef
> base::Callback<void(int device_id, unsigned int code, bool down)>
> On 2015/01/26 22:18:25, alexst wrote:
> > My brain wants more structure here. 
> > 
> > You are passing things individually in one place, but as a struct in
another.
> I
> > know there is an interface down the road that replaces these, but similarly,
> it
> > splits between the two styles.
> > 
> > I gather it's a callback param limit thing, but since we are bumping against
> it
> > in one place, would you hate me if I asked we moved all of these to structs?
> 
> We could remove the structs once the 7 argument limit is removed. There is a
> patch for this in review.
> 
> Why make more structs just to remove them?

I meant we could've kept the structs, but my main concern is consistency. Since
you don't like them, I'm okay with a TODO to remove them once the other patch
lands.

Otherwise lgtm.

[spang, 2015-01-27 21:31:54]

https://codereview.chromium.org/872883003/diff/20001/ui/events/ozone/evdev/ev...
File ui/events/ozone/evdev/event_dispatch_callback.h (right):

https://codereview.chromium.org/872883003/diff/20001/ui/events/ozone/evdev/ev...
ui/events/ozone/evdev/event_dispatch_callback.h:29: typedef
base::Callback<void(int device_id, unsigned int code, bool down)>
On 2015/01/26 23:01:21, spang wrote:
> On 2015/01/26 22:18:25, alexst wrote:
> > My brain wants more structure here. 
> > 
> > You are passing things individually in one place, but as a struct in
another.
> I
> > know there is an interface down the road that replaces these, but similarly,
> it
> > splits between the two styles.
> > 
> > I gather it's a callback param limit thing, but since we are bumping against
> it
> > in one place, would you hate me if I asked we moved all of these to structs?
> 
> We could remove the structs once the 7 argument limit is removed. There is a
> patch for this in review.
> 
> Why make more structs just to remove them?

Done.

[spang, 2015-01-27 22:39:05]

The CQ bit was checked by spang@chromium.org

[commit-bot: I haz the power, 2015-01-27 22:40:09]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/872883003/40001

[commit-bot: I haz the power, 2015-01-27 23:30:46]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-01-27 23:32:44]

Patchset 3 (id:??) landed as
https://crrev.com/7b5b72bda2a843c48cb268181382d65f668c78ba
Cr-Commit-Position: refs/heads/master@{#313397}