Add casts for EGLNativeDisplayType to XDisplay to use X functions with OZONE.

Add casts for EGLNativeDisplayType to XDisplay to use X functions with OZONE.

OZONE build defines EGLNativeDisplayType like intptr_t .
It requires reinterpret custs to be used with X functions.

BUG=361137

[ostap, 2014-11-11 20:35:20]

sl.ostapenko@samsung.com changed reviewers:
+ alexst@chromium.org, piman@chromium.org

[alexst (slow to review), 2014-11-11 20:42:44]

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

[piman, 2014-11-11 21:23:27]

lgtm

[spang, 2014-11-11 21:26:32]

What are you doing that gets you gl_surface_x11.cc with use_ozone=1 ?

This isn't expected to work, and will cause a bunch of symbol conflicts.

[ostap, 2014-11-11 22:28:08]

On 2014/11/11 21:26:32, spang wrote:
> What are you doing that gets you gl_surface_x11.cc with use_ozone=1 ?
> 
> This isn't expected to work, and will cause a bunch of symbol conflicts.

Actually, why? What would conflict?
The same build on linux works with --use-gl=egl and --use-gl=desktop

I'm trying to do ozone build with x11 working. Going through all things that
break when I add both -Duse_ozone=1 and -Duse_x11=1 .

[spang, 2014-11-11 23:04:57]

On 2014/11/11 22:28:08, ostap wrote:
> On 2014/11/11 21:26:32, spang wrote:
> > What are you doing that gets you gl_surface_x11.cc with use_ozone=1 ?
> > 
> > This isn't expected to work, and will cause a bunch of symbol conflicts.
> 
> Actually, why? What would conflict?

Quite a few things will conflict. For example, the function you are changing:

ui/gl/gl_surface_ozone.cc:EGLNativeDisplayType
GetPlatformDefaultEGLNativeDisplay() {
ui/gl/gl_surface_x11.cc:EGLNativeDisplayType
GetPlatformDefaultEGLNativeDisplay() {

Both objects will be built in your config, causing unpredictable linker behavior
at best.

This isn't a valid config (yet).

> The same build on linux works with --use-gl=egl and --use-gl=desktop
> 
> I'm trying to do ozone build with x11 working. Going through all things that
> break when I add both -Duse_ozone=1 and -Duse_x11=1 .

[ostap, 2014-11-11 23:20:37]

On 2014/11/11 23:04:57, spang wrote:
> On 2014/11/11 22:28:08, ostap wrote:
> > On 2014/11/11 21:26:32, spang wrote:
> This isn't a valid config (yet).

Yes, I know this.
That is what I actually trying to fix.

With ozone ::Display, NativeEvent and NativeCursor are defined differently.
That's what I'm fixing at the 1st stage.

[spang, 2014-11-11 23:29:14]

On 2014/11/11 23:20:37, ostap wrote:
> On 2014/11/11 23:04:57, spang wrote:
> > On 2014/11/11 22:28:08, ostap wrote:
> > > On 2014/11/11 21:26:32, spang wrote:
> > This isn't a valid config (yet).
> 
> Yes, I know this.
> That is what I actually trying to fix.
> 
> With ozone ::Display, NativeEvent and NativeCursor are defined differently.
> That's what I'm fixing at the 1st stage.

Right, but adding ozone ifdefs with casts in the x11 code isn't how we're hoping
to deal with those issues. And we definitely can't have multiple definitions of
functions in the same build.

If you could send us a note at ozone-dev@chromium.org so we can coordinate this
effort, that would be helpful. This will avoid duplicated effort since we're
also looking into bringing up an x11 ozone build.

[kalyank, 2014-11-11 23:31:49]

On 2014/11/11 23:29:14, spang wrote:
> On 2014/11/11 23:20:37, ostap wrote:
> > On 2014/11/11 23:04:57, spang wrote:
> > > On 2014/11/11 22:28:08, ostap wrote:
> > > > On 2014/11/11 21:26:32, spang wrote:
> > > This isn't a valid config (yet).
> > 
> > Yes, I know this.
> > That is what I actually trying to fix.
> > 
> > With ozone ::Display, NativeEvent and NativeCursor are defined differently.
> > That's what I'm fixing at the 1st stage.
> 
> Right, but adding ozone ifdefs with casts in the x11 code isn't how we're
hoping
> to deal with those issues. And we definitely can't have multiple definitions
of
> functions in the same build.
> 
> If you could send us a note at mailto:ozone-dev@chromium.org so we can
coordinate this
> effort, that would be helpful. This will avoid duplicated effort since we're
> also looking into bringing up an x11 ozone build.

Ideally, on Ozone we will be using GLSurfaceOzoneEGL and have a proper
ui/ozone/public/surface_factory_ozone.h implementation for x11.

[ostap, 2014-11-12 16:11:42]

On 2014/11/11 23:31:49, kalyank wrote:
> On 2014/11/11 23:29:14, spang wrote:
> > On 2014/11/11 23:20:37, ostap wrote:
> > > On 2014/11/11 23:04:57, spang wrote:
> > > > On 2014/11/11 22:28:08, ostap wrote:
> > > > > On 2014/11/11 21:26:32, spang wrote:
> Ideally, on Ozone we will be using GLSurfaceOzoneEGL and have a proper
> ui/ozone/public/surface_factory_ozone.h implementation for x11.

Hi Kalyan,

So, I shouldn't enable both use_x11 and use_ozone, but just add another ozone
platform, right?
Should it be implemented for egl only, or it would be nice to have glx
implementation also?

[kalyank, 2014-11-12 16:43:44]

> Should it be implemented for egl only, or it would be nice to have glx
> implementation also?

Hi Ostap,

>>I shouldn't enable both use_x11 and use_ozone
Ya, the goal of Ozone is to do a run time binding of the platform rather than a
compile time. So, you shouldn't enable
use_x11.If you want to look at an example implementation, check ozone
implementations under ui/ozone/platform or
https://github.com/01org/ozone-wayland

Reg a new X11 ozone platform:
You should check this CL https://codereview.chromium.org/705293003/ Their is
already ongoing work for this. 

Reg GLX support @piman, @spang are the right people to talk to. 

You should probably send a mail to Ozone mailing list as Spang suggested or
start the discussion in 361137 reg this. If you prefer IRC #chromium or
#ozone-wayland in freenode. 
Chromium for Desktop builds  still need
https://codereview.chromium.org/36953002/ to get resolved. I hope to put up a CL
for this by end of this week.

[ostap, 2014-11-12 18:04:53]

On 2014/11/12 16:43:44, kalyank wrote:
> > Should it be implemented for egl only, or it would be nice to have glx
> > implementation also?
> 
> Hi Ostap,
> 
> >>I shouldn't enable both use_x11 and use_ozone
> Ya, the goal of Ozone is to do a run time binding of the platform rather than
a
> compile time. So, you shouldn't enable
> use_x11.If you want to look at an example implementation, check ozone
> implementations under ui/ozone/platform or
> https://github.com/01org/ozone-wayland

Thanks. I think I've got an idea.