mojo: Convert gles2 to the new thunking system

mojo: Convert gles2 to the new thunking system.

BUG=386799
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=286365

[viettrungluu, 2014-07-28 22:11:05]

https://codereview.chromium.org/413303002/diff/60001/mojo/public/platform/nat...
File mojo/public/platform/native/gles2_thunks.h (right):

https://codereview.chromium.org/413303002/diff/60001/mojo/public/platform/nat...
mojo/public/platform/native/gles2_thunks.h:37: #include
"mojo/public/c/gles2/gles2_call_visitor_autogen.h"
Is this resilient to changes? Is autogen going to generate things in a given
order (and then only add things to the end, and not make any changes)?

Even if so, how would you add any calls "manually"?

[Elliot Glaysher, 2014-07-28 22:21:27]

https://codereview.chromium.org/413303002/diff/60001/mojo/public/platform/nat...
File mojo/public/platform/native/gles2_thunks.h (right):

https://codereview.chromium.org/413303002/diff/60001/mojo/public/platform/nat...
mojo/public/platform/native/gles2_thunks.h:37: #include
"mojo/public/c/gles2/gles2_call_visitor_autogen.h"
On 2014/07/28 22:11:05, viettrungluu wrote:
> Is this resilient to changes? Is autogen going to generate things in a given
> order (and then only add things to the end, and not make any changes)?
> 
> Even if so, how would you add any calls "manually"?

The functions added from gles2_call_visitor_autogen.h are a frozen API and
changing it would also break PPAPI (if I understand piman@ correctly).

(Somebody (jamesr@?) is working on removing the leaked interface above. The rest
of the interface here is solidifying.)

[Elliot Glaysher, 2014-07-29 00:03:16]

trung: so I've split the GLES2 implementation from the control functions in this
update. ptal.

[viettrungluu, 2014-07-29 00:27:06]

LGTM w/nits.

https://codereview.chromium.org/413303002/diff/100001/mojo/public/platform/na...
File mojo/public/platform/native/gles2_thunks.cc (right):

https://codereview.chromium.org/413303002/diff/100001/mojo/public/platform/na...
mojo/public/platform/native/gles2_thunks.cc:65: // Call this function by looking
If only I knew what this comment (which you copied from an existing file) was
meant to say.

https://codereview.chromium.org/413303002/diff/100001/mojo/public/platform/na...
mojo/public/platform/native/gles2_thunks.cc:68: #define THUNK_EXPORT
__declspec(dllexport)
Could you factor THUNK_EXPORT out into a header file
(.../platform/native/thunk_export.h?).

https://codereview.chromium.org/413303002/diff/100001/mojo/public/platform/na...
mojo/public/platform/native/gles2_thunks.cc:75: if (gles2_control_thunks->size
>= sizeof(g_control_thunks)) {
nit (here and below): Mostly, we omit the braces when the block is a single
line.

https://codereview.chromium.org/413303002/diff/100001/mojo/public/platform/na...
File mojo/public/platform/native/gles2_thunks.h (right):

https://codereview.chromium.org/413303002/diff/100001/mojo/public/platform/na...
mojo/public/platform/native/gles2_thunks.h:13: // DSO to theose of the embedder.
s/theose/those/

https://codereview.chromium.org/413303002/diff/100001/mojo/public/platform/na...
mojo/public/platform/native/gles2_thunks.h:22: void (*GLES2Terminate)(void);
This is actually C++ here; no extra "void" needed.

https://codereview.chromium.org/413303002/diff/100001/mojo/public/platform/na...
mojo/public/platform/native/gles2_thunks.h:29: void (*GLES2SwapBuffers)(void);
"

https://codereview.chromium.org/413303002/diff/100001/mojo/public/platform/na...
mojo/public/platform/native/gles2_thunks.h:80: // Intended to be called from the
embedder to get the embeder's implementation
s/embeder/embedder/

https://codereview.chromium.org/413303002/diff/100001/mojo/shell/in_process_d...
File mojo/shell/in_process_dynamic_service_runner.cc (right):

https://codereview.chromium.org/413303002/diff/100001/mojo/shell/in_process_d...
mojo/shell/in_process_dynamic_service_runner.cc:127: << "doesn't have
MojoSetGLES2ImplThunks.";
No need for extra <<.

[Elliot Glaysher, 2014-07-29 22:02:29]

The CQ bit was checked by erg@chromium.org

[commit-bot: I haz the power, 2014-07-29 22:04:27]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/erg@chromium.org/413303002/140001

[commit-bot: I haz the power, 2014-07-30 00:46:29]

Change committed as 286365

[tapted, 2014-07-30 03:09:07]

A revert of this CL has been created in
https://codereview.chromium.org/425193002/ by tapted@chromium.org.

The reason for reverting is: Suspected for breaking official mac builder. Link:

http://build.chromium.org/p/chromium.chrome/builders/Google%20Chrome%20Mac/bu...

Error is:

FAILED:  /* snip */ python ../tools/isolate_driver.py check --isolated
"../out/Release/mojo_apps_js_unittests.isolated" --isolate
"mojo_apps_js_unittests.isolate" /*snip*/
Traceback (most recent call last):
  File
"/Volumes/data/b/build/slave/google-chrome-rel-mac/build/src/tools/swarming_client/utils/file_path.py",
line 488, in fix_native_path_case
    os.path.join(native_case_path, raw_part))
OSError: File
/Volumes/data/b/build/slave/google-chrome-rel-mac/build/src/out/Release/libmojo_gles2.dylib
doesn't exist
.

[skobes, 2014-07-30 03:44:05]

FYI, in addition to the Mac compile failure, this also broke
mojo_view_manager_unittests on various Linux Clang bots.

http://build.chromium.org/p/chromium.linux/builders/Linux%20Clang%20%28dbg%29...

../../mojo/public/platform/native/gles2_thunks.cc:17: void
MojoGLES2Initialize(const MojoAsyncWaiter *): Assertion
`g_control_thunks.GLES2Initialize' failed.