Add PPB_Fullscreen;0.5.

Add PPB_Fullscreen_Dev;0.5. Keep 0.4 for backwards compatiblity and point it
to PPB_FlashFullscreen. The new implementation is based on
http://codereview.chromium.org/7714017/ with some bug fixes.
Update header comments.

Main API differences between the old and the new implementation:
- transition from fullscreen is now asynchronous and ends at DidChangeView
just like transition to fullscreen; graphics devices cannot be bound during
the transition.
- when switching to/from fullscreen 3D resources no longer need to be
re-created.
- transitions to fullscreen are only possible when processing user user gestures.
- transition to fullscreen results in 2 DidChangeViews, one for moving the
plugin to the middle of the window and one for stretching the window and
placing the plugin in the middle of the screen.
- the size of the plugin is not changed when going to/from fullscreen.

Testing:
- Mapped ppapi_tests:test_fullscreen to ppapi_tests:test_flash_fullscreen.
- Updated test_fullscreen to work with the new implementation. To be testable
automatically this needs enhancements to the testing infrastructure
for generating user gestures. For now marked the test as DISABLED.
- Disabled NaCl's ppapi_ppb_fullscreen_browser_test for the same reasons
as above.
- To re-enable both tests, we will first need to add user gesture capabilites
to PPB_Testing.
- Build 0.4 ppapi_test:test_fullscreen and ran this out of process and in
process with the newly build revision of chrome to verify backwards
compatability.
- In a separate CL, will update NaCl's ppapi_ppb_fullscreen_browser_test to work
with the new implementation, for now only manually.

BUG=41780
TEST=see above
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=102888

[polina, 2011-09-09 00:38:16]

This is a draft. It is passing with a NaCl-based test for PPB_Fullscreen_Dev, so
I have not messed up the existing code. As for new PPB_Fullscreen,  I am still
working out the ppapi_tests setup here: http://codereview.chromium.org/7860005,
so I can use that test to verify the changes in this CL. I am going to send out
NaCl proxy and test for PPB_Fullscreen in a separate CL.

[piman, 2011-09-09 01:33:44]

http://codereview.chromium.org/7826017/diff/19001/ppapi/api/dev/ppb_fullscree...
File ppapi/api/dev/ppb_fullscreen_dev.idl (right):

http://codereview.chromium.org/7826017/diff/19001/ppapi/api/dev/ppb_fullscree...
ppapi/api/dev/ppb_fullscreen_dev.idl:44: PP_Bool GetScreenSize(
Maybe we can remove this function from the Dev interface because it has the
exact same implementation (and purpose) as the stable one.

http://codereview.chromium.org/7826017/diff/19001/ppapi/proxy/ppp_instance_pr...
File ppapi/proxy/ppp_instance_proxy.cc (right):

http://codereview.chromium.org/7826017/diff/19001/ppapi/proxy/ppp_instance_pr...
ppapi/proxy/ppp_instance_proxy.cc:62: PP_Bool fullscreen_either = ((fullscreen
== PP_TRUE ||
You need to send both flags separately if you want to implement IsFullscreen
consistently for in-process and out-of-process.
An alternative is to make IsFullscreen return true if we're in either mode, in
which case you would change the in-process implementation to report that, and
you could then remove IsFullscreen from the Dev interface because it's redundant
with the non-Dev.

http://codereview.chromium.org/7826017/diff/19001/webkit/plugins/ppapi/ppapi_...
File webkit/plugins/ppapi/ppapi_webplugin_impl.cc (right):

http://codereview.chromium.org/7826017/diff/19001/webkit/plugins/ppapi/ppapi_...
webkit/plugins/ppapi/ppapi_webplugin_impl.cc:123: instance_->Paint(canvas,
plugin_rect_, rect);
Maybe I'm confused about the implementation, but shouldn't Paint be called here
if in the "new" fullscreen mode? If not, when/how does Paint get called? Same
with ViewChanged and HandleInputEvent below.

[jeremya, 2011-09-09 17:45:35]

http://codereview.chromium.org/7826017/diff/19001/webkit/plugins/ppapi/ppapi_...
File webkit/plugins/ppapi/ppapi_plugin_instance.cc (right):

http://codereview.chromium.org/7826017/diff/19001/webkit/plugins/ppapi/ppapi_...
webkit/plugins/ppapi/ppapi_plugin_instance.cc:919: BindGraphics(pp_instance(),
0);
Is it necessary to unbind the graphics context here?

http://codereview.chromium.org/7826017/diff/19001/webkit/plugins/ppapi/ppapi_...
File webkit/plugins/ppapi/ppapi_webplugin_impl.cc (right):

http://codereview.chromium.org/7826017/diff/19001/webkit/plugins/ppapi/ppapi_...
webkit/plugins/ppapi/ppapi_webplugin_impl.cc:123: instance_->Paint(canvas,
plugin_rect_, rect);
On 2011/09/09 01:33:44, piman wrote:
> Maybe I'm confused about the implementation, but shouldn't Paint be called
here
> if in the "new" fullscreen mode? If not, when/how does Paint get called? Same
> with ViewChanged and HandleInputEvent below.

I think you're right -- in the old fullscreen mode, the widget did its own
painting, event handling, etc.

[jeremya, 2011-09-23 20:04:28]

I just tried to patch this in to a local client, and it doesn't apply:

$ git cl patch 7826017     
error: ppapi/tests/test_flash_fullscreen.cc: does not exist in index
error: ppapi/tests/test_flash_fullscreen.h: does not exist in index
Failed to apply the patch

[polina, 2011-09-26 21:57:00]

New version uploaded. Please take a look. The approach and the surrounding code
have changed significantly since the 1st patch set, so the relative diff might
not be very useful.

http://codereview.chromium.org/7826017/diff/19001/ppapi/api/dev/ppb_fullscree...
File ppapi/api/dev/ppb_fullscreen_dev.idl (right):

http://codereview.chromium.org/7826017/diff/19001/ppapi/api/dev/ppb_fullscree...
ppapi/api/dev/ppb_fullscreen_dev.idl:44: PP_Bool GetScreenSize(
On 2011/09/09 01:33:44, piman wrote:
> Maybe we can remove this function from the Dev interface because it has the
> exact same implementation (and purpose) as the stable one.

I don't think we should since the double-interface situation is temporary. This
will require combined usage of the two interfaces, which can complicate things.

http://codereview.chromium.org/7826017/diff/19001/ppapi/proxy/ppp_instance_pr...
File ppapi/proxy/ppp_instance_proxy.cc (right):

http://codereview.chromium.org/7826017/diff/19001/ppapi/proxy/ppp_instance_pr...
ppapi/proxy/ppp_instance_proxy.cc:62: PP_Bool fullscreen_either = ((fullscreen
== PP_TRUE ||
On 2011/09/09 01:33:44, piman wrote:
> You need to send both flags separately if you want to implement IsFullscreen
> consistently for in-process and out-of-process.
> An alternative is to make IsFullscreen return true if we're in either mode, in
> which case you would change the in-process implementation to report that, and
> you could then remove IsFullscreen from the Dev interface because it's
redundant
> with the non-Dev.

Done.

http://codereview.chromium.org/7826017/diff/19001/webkit/plugins/ppapi/ppapi_...
File webkit/plugins/ppapi/ppapi_plugin_instance.cc (right):

http://codereview.chromium.org/7826017/diff/19001/webkit/plugins/ppapi/ppapi_...
webkit/plugins/ppapi/ppapi_plugin_instance.cc:919: BindGraphics(pp_instance(),
0);
On 2011/09/09 17:45:35, jeremya wrote:
> Is it necessary to unbind the graphics context here?

That's what the interface documentation in the header says should happen and how
the old implementation works. This also seems to be consistent with not allowing
device binding during transitions.

That said, I do have a concern that this is inconsistent with what a user needs
to do on resize. No rebinding is necessary in that case, just a ResizeBuffers
call. I would be open to making this consistent. For example, we could unbind
and rebind for the user implicitly, so that he she only needs to resize to the
fullscreen in DidChangeView.

http://codereview.chromium.org/7826017/diff/19001/webkit/plugins/ppapi/ppapi_...
File webkit/plugins/ppapi/ppapi_webplugin_impl.cc (right):

http://codereview.chromium.org/7826017/diff/19001/webkit/plugins/ppapi/ppapi_...
webkit/plugins/ppapi/ppapi_webplugin_impl.cc:123: instance_->Paint(canvas,
plugin_rect_, rect);
On 2011/09/09 17:45:35, jeremya wrote:
> On 2011/09/09 01:33:44, piman wrote:
> > Maybe I'm confused about the implementation, but shouldn't Paint be called
> here
> > if in the "new" fullscreen mode? If not, when/how does Paint get called?
Same
> > with ViewChanged and HandleInputEvent below.
> 
> I think you're right -- in the old fullscreen mode, the widget did its own
> painting, event handling, etc.

I did not have the full background on the differences between the old and the
new implementation. The code should make more sense now.

[polina, 2011-09-26 22:12:36]

http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscree...
File ppapi/api/dev/ppb_fullscreen_dev.idl (right):

http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscree...
ppapi/api/dev/ppb_fullscreen_dev.idl:37: * Plugin size will not be affected.
I am updating the header comments in a different patch set to draw your
attention to how different the new fullscreen implementation is from what was
expected based on the old one. The extensive changes I had to make to the test
also demonstrate that.

I propose that we consider suppressing the first DidChangeView for fullscreen
transitions. I think the 2-step nature of the transition is an implementation
detail that should be hidden from the developers.

I would also like to add a way to determine if SetFullscreen was called while
processing a user gesture, so we can return PP_FALSE otherwise. Currently
PP_TRUE is always returned by SetFullscreen. I have not yet figured out how to
check that. Please let me know if you know of a quick way to do so.

[polina, 2011-09-26 22:24:10]

http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscree...
File ppapi/api/dev/ppb_fullscreen_dev.idl (right):

http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscree...
ppapi/api/dev/ppb_fullscreen_dev.idl:37: * Plugin size will not be affected.
Also, please not that the plugin is not resized to occupy the entire screen.
What approach do we recommend developers use to address that?

[piman, 2011-09-26 23:01:03]

http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscree...
File ppapi/api/dev/ppb_fullscreen_dev.idl (right):

http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscree...
ppapi/api/dev/ppb_fullscreen_dev.idl:37: * Plugin size will not be affected.
On 2011/09/26 22:24:10, polina wrote:
> Also, please not that the plugin is not resized to occupy the entire screen.
> What approach do we recommend developers use to address that?

I don't understand this... How can a plugin deal with full-screen if the
instance isn't at the size of the full screen?

[brettw, 2011-09-27 00:04:20]

I'm OK with the double interface for now.

LGTM

http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen.cc
File ppapi/tests/test_fullscreen.cc (right):

http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen...
ppapi/tests/test_fullscreen.cc:34: error_(""),
Just use error_() (the "" version requires more work for no benefit).

http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen...
ppapi/tests/test_fullscreen.cc:64: error_ = "";
error_.clear()?

http://codereview.chromium.org/7826017/diff/50041/ppapi/thunk/ppb_instance_api.h
File ppapi/thunk/ppb_instance_api.h (right):

http://codereview.chromium.org/7826017/diff/50041/ppapi/thunk/ppb_instance_ap...
ppapi/thunk/ppb_instance_api.h:60: 
Extra blank?

http://codereview.chromium.org/7826017/diff/50041/webkit/plugins/ppapi/ppapi_...
File webkit/plugins/ppapi/ppapi_plugin_instance.cc (right):

http://codereview.chromium.org/7826017/diff/50041/webkit/plugins/ppapi/ppapi_...
webkit/plugins/ppapi/ppapi_plugin_instance.cc:940: BindGraphics(pp_instance(),
0);
FYI I'm guessing we'll want to remove this and keep the old device bound. We can
discuss this later. I think the "everything will be unbound" in the spec was
driven by implementation details in the old code.

[polina, 2011-09-27 00:49:08]

http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscree...
File ppapi/api/dev/ppb_fullscreen_dev.idl (right):

http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscree...
ppapi/api/dev/ppb_fullscreen_dev.idl:37: * Plugin size will not be affected.
On 2011/09/26 23:01:03, piman wrote:
> On 2011/09/26 22:24:10, polina wrote:
> > Also, please not that the plugin is not resized to occupy the entire screen.
> > What approach do we recommend developers use to address that?
> 
> I don't understand this... How can a plugin deal with full-screen if the
> instance isn't at the size of the full screen?

I was surprised by this behavior myself. In his JS demo here:
http://www.corp.google.com/~jeremya/fullscreen.html, Jeremy used
 
<style>
    <...>
    :-webkit-full-screen {
        width: 100%; height: 100%;
    }
</style>

I am not aware of a way to resize the plugin to larger dimensions from Pepper
C/C++. And I would prefer it if developers did not have to worry about this in
either JS or C/C++. I think we should do it for them. Jeremy can comment on why
he implemented things this way, and if he is planning on changing this behavior.
If not, we should consider addressing it in the Pepper layer although it is
undesirable for us not to match JS behavior.

http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen.cc
File ppapi/tests/test_fullscreen.cc (right):

http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen...
ppapi/tests/test_fullscreen.cc:34: error_(""),
On 2011/09/27 00:04:20, brettw wrote:
> Just use error_() (the "" version requires more work for no benefit).

Done.

http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen...
ppapi/tests/test_fullscreen.cc:64: error_ = "";
On 2011/09/27 00:04:20, brettw wrote:
> error_.clear()?

Done.

http://codereview.chromium.org/7826017/diff/50041/ppapi/thunk/ppb_instance_api.h
File ppapi/thunk/ppb_instance_api.h (right):

http://codereview.chromium.org/7826017/diff/50041/ppapi/thunk/ppb_instance_ap...
ppapi/thunk/ppb_instance_api.h:60: 
On 2011/09/27 00:04:20, brettw wrote:
> Extra blank?

Done.

http://codereview.chromium.org/7826017/diff/50041/webkit/plugins/ppapi/ppapi_...
File webkit/plugins/ppapi/ppapi_plugin_instance.cc (right):

http://codereview.chromium.org/7826017/diff/50041/webkit/plugins/ppapi/ppapi_...
webkit/plugins/ppapi/ppapi_plugin_instance.cc:940: BindGraphics(pp_instance(),
0);
On 2011/09/27 00:04:20, brettw wrote:
> FYI I'm guessing we'll want to remove this and keep the old device bound. We
can
> discuss this later. I think the "everything will be unbound" in the spec was
> driven by implementation details in the old code.

Please see my response to question Jeremy had about this 1st patch set. This is
one of the issues that I wanted to have an API review for.

[piman, 2011-09-27 00:52:39]

On Mon, Sep 26, 2011 at 5:49 PM, <polina@google.com> wrote:

>
> http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
>
api/dev/ppb_fullscreen_dev.idl<http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscreen_dev.idl>
> File ppapi/api/dev/ppb_fullscreen_**dev.idl (right):
>
> http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
>
api/dev/ppb_fullscreen_dev.**idl#newcode37<http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscreen_dev.idl#newcode37>
> ppapi/api/dev/ppb_fullscreen_**dev.idl:37: * Plugin size will not be
> affected.
> On 2011/09/26 23:01:03, piman wrote:
>
>> On 2011/09/26 22:24:10, polina wrote:
>> > Also, please not that the plugin is not resized to occupy the entire
>>
> screen.
>
>> > What approach do we recommend developers use to address that?
>>
>
>  I don't understand this... How can a plugin deal with full-screen if
>>
> the
>
>> instance isn't at the size of the full screen?
>>
>
> I was surprised by this behavior myself. In his JS demo here:
>
http://www.corp.google.com/~**jeremya/fullscreen.html<http://www.corp.google....,
> Jeremy used
>
> <style>
>    <...>
>    :-webkit-full-screen {
>        width: 100%; height: 100%;
>    }
> </style>
>
> I am not aware of a way to resize the plugin to larger dimensions from
> Pepper C/C++. And I would prefer it if developers did not have to worry
> about this in either JS or C/C++. I think we should do it for them.
> Jeremy can comment on why he implemented things this way, and if he is
> planning on changing this behavior. If not, we should consider
> addressing it in the Pepper layer although it is undesirable for us not
> to match JS behavior.


FYI Flash needs this. Behind-the-scenes "transparent" scaling is not enough.

Antoine


>
>
> http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
>
tests/test_fullscreen.cc<http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen.cc>
> File ppapi/tests/test_fullscreen.cc (right):
>
> http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
>
tests/test_fullscreen.cc#**newcode34<http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen.cc#newcode34>
> ppapi/tests/test_fullscreen.**cc:34: error_(""),
> On 2011/09/27 00:04:20, brettw wrote:
>
>> Just use error_() (the "" version requires more work for no benefit).
>>
>
> Done.
>
>
> http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
>
tests/test_fullscreen.cc#**newcode64<http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen.cc#newcode64>
> ppapi/tests/test_fullscreen.**cc:64: error_ = "";
> On 2011/09/27 00:04:20, brettw wrote:
>
>> error_.clear()?
>>
>
> Done.
>
>
> http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
>
thunk/ppb_instance_api.h<http://codereview.chromium.org/7826017/diff/50041/ppapi/thunk/ppb_instance_api.h>
> File ppapi/thunk/ppb_instance_api.h (right):
>
> http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
>
thunk/ppb_instance_api.h#**newcode60<http://codereview.chromium.org/7826017/diff/50041/ppapi/thunk/ppb_instance_api.h#newcode60>
> ppapi/thunk/ppb_instance_api.**h:60:
> On 2011/09/27 00:04:20, brettw wrote:
>
>> Extra blank?
>>
>
> Done.
>
>
> http://codereview.chromium.**org/7826017/diff/50041/webkit/**
>
plugins/ppapi/ppapi_plugin_**instance.cc<http://codereview.chromium.org/7826017/diff/50041/webkit/plugins/ppapi/ppapi_plugin_instance.cc>
> File webkit/plugins/ppapi/ppapi_**plugin_instance.cc (right):
>
> http://codereview.chromium.**org/7826017/diff/50041/webkit/**
>
plugins/ppapi/ppapi_plugin_**instance.cc#newcode940<http://codereview.chromium.org/7826017/diff/50041/webkit/plugins/ppapi/ppapi_plugin_instance.cc#newcode940>
> webkit/plugins/ppapi/ppapi_**plugin_instance.cc:940:
> BindGraphics(pp_instance(), 0);
> On 2011/09/27 00:04:20, brettw wrote:
>
>> FYI I'm guessing we'll want to remove this and keep the old device
>>
> bound. We can
>
>> discuss this later. I think the "everything will be unbound" in the
>>
> spec was
>
>> driven by implementation details in the old code.
>>
>
> Please see my response to question Jeremy had about this 1st patch set.
> This is one of the issues that I wanted to have an API review for.
>
>
>
http://codereview.chromium.**org/7826017/<http://codereview.chromium.org/7826...
>

[polina, 2011-09-27 01:06:13]

That's useful to know. Since Flash has a back-up implementation, my plan is to
get this CL as-is and address all the concerns we have for Flash and otherwise
in later CLs.

Thanks,
P.

On 2011/09/27 00:52:39, piman wrote:
> On Mon, Sep 26, 2011 at 5:49 PM, <mailto:polina@google.com> wrote:
> 
> >
> > http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
> >
>
api/dev/ppb_fullscreen_dev.idl<http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscreen_dev.idl>
> > File ppapi/api/dev/ppb_fullscreen_**dev.idl (right):
> >
> > http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
> >
>
api/dev/ppb_fullscreen_dev.**idl#newcode37<http://codereview.chromium.org/7826017/diff/50041/ppapi/api/dev/ppb_fullscreen_dev.idl#newcode37>
> > ppapi/api/dev/ppb_fullscreen_**dev.idl:37: * Plugin size will not be
> > affected.
> > On 2011/09/26 23:01:03, piman wrote:
> >
> >> On 2011/09/26 22:24:10, polina wrote:
> >> > Also, please not that the plugin is not resized to occupy the entire
> >>
> > screen.
> >
> >> > What approach do we recommend developers use to address that?
> >>
> >
> >  I don't understand this... How can a plugin deal with full-screen if
> >>
> > the
> >
> >> instance isn't at the size of the full screen?
> >>
> >
> > I was surprised by this behavior myself. In his JS demo here:
> >
>
http://www.corp.google.com/%7E**jeremya/fullscreen.html%3Chttp://www.corp.goo...>,
> > Jeremy used
> >
> > <style>
> >    <...>
> >    :-webkit-full-screen {
> >        width: 100%; height: 100%;
> >    }
> > </style>
> >
> > I am not aware of a way to resize the plugin to larger dimensions from
> > Pepper C/C++. And I would prefer it if developers did not have to worry
> > about this in either JS or C/C++. I think we should do it for them.
> > Jeremy can comment on why he implemented things this way, and if he is
> > planning on changing this behavior. If not, we should consider
> > addressing it in the Pepper layer although it is undesirable for us not
> > to match JS behavior.
> 
> 
> FYI Flash needs this. Behind-the-scenes "transparent" scaling is not enough.
> 
> Antoine
> 
> 
> >
> >
> > http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
> >
>
tests/test_fullscreen.cc<http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen.cc>
> > File ppapi/tests/test_fullscreen.cc (right):
> >
> > http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
> >
>
tests/test_fullscreen.cc#**newcode34<http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen.cc#newcode34>
> > ppapi/tests/test_fullscreen.**cc:34: error_(""),
> > On 2011/09/27 00:04:20, brettw wrote:
> >
> >> Just use error_() (the "" version requires more work for no benefit).
> >>
> >
> > Done.
> >
> >
> > http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
> >
>
tests/test_fullscreen.cc#**newcode64<http://codereview.chromium.org/7826017/diff/50041/ppapi/tests/test_fullscreen.cc#newcode64>
> > ppapi/tests/test_fullscreen.**cc:64: error_ = "";
> > On 2011/09/27 00:04:20, brettw wrote:
> >
> >> error_.clear()?
> >>
> >
> > Done.
> >
> >
> > http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
> >
>
thunk/ppb_instance_api.h<http://codereview.chromium.org/7826017/diff/50041/ppapi/thunk/ppb_instance_api.h>
> > File ppapi/thunk/ppb_instance_api.h (right):
> >
> > http://codereview.chromium.**org/7826017/diff/50041/ppapi/**
> >
>
thunk/ppb_instance_api.h#**newcode60<http://codereview.chromium.org/7826017/diff/50041/ppapi/thunk/ppb_instance_api.h#newcode60>
> > ppapi/thunk/ppb_instance_api.**h:60:
> > On 2011/09/27 00:04:20, brettw wrote:
> >
> >> Extra blank?
> >>
> >
> > Done.
> >
> >
> > http://codereview.chromium.**org/7826017/diff/50041/webkit/**
> >
>
plugins/ppapi/ppapi_plugin_**instance.cc<http://codereview.chromium.org/7826017/diff/50041/webkit/plugins/ppapi/ppapi_plugin_instance.cc>
> > File webkit/plugins/ppapi/ppapi_**plugin_instance.cc (right):
> >
> > http://codereview.chromium.**org/7826017/diff/50041/webkit/**
> >
>
plugins/ppapi/ppapi_plugin_**instance.cc#newcode940<http://codereview.chromium.org/7826017/diff/50041/webkit/plugins/ppapi/ppapi_plugin_instance.cc#newcode940>
> > webkit/plugins/ppapi/ppapi_**plugin_instance.cc:940:
> > BindGraphics(pp_instance(), 0);
> > On 2011/09/27 00:04:20, brettw wrote:
> >
> >> FYI I'm guessing we'll want to remove this and keep the old device
> >>
> > bound. We can
> >
> >> discuss this later. I think the "everything will be unbound" in the
> >>
> > spec was
> >
> >> driven by implementation details in the old code.
> >>
> >
> > Please see my response to question Jeremy had about this 1st patch set.
> > This is one of the issues that I wanted to have an API review for.
> >
> >
> >
>
http://codereview.chromium.**org/7826017/%3Chttp://codereview.chromium.org/78...>
> >