Skia DEPS roll to 3147

Skia DEPS roll to 3147
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=121451

[epoger, 2012-02-07 17:07:29]

Includes chromium changes needed for r3147, taken from
https://chromiumcodereview.appspot.com/9341003/ .

I'm going to wait until Wednesday morning to run this through the trybots,
because the trybots are in disarray today.  I will request review once the
trybots complete on Wednesday.

[epoger, 2012-02-08 12:52:09]

out to the trybots (although I don't know if they're working yet)

control group also out to the trybots ; see
https://chromiumcodereview.appspot.com/9348055

[epoger, 2012-02-08 14:05:30]

On 2012/02/08 12:52:09, epoger wrote:
> out to the trybots (although I don't know if they're working yet)

Looking at http://build.chromium.org/p/tryserver.chromium/waterfall , it seems
that no trybots have run since 1:41 AM Pacific time.  So they are still dead. 
I'll try again tomorrow...

> 
> control group also out to the trybots ; see
> https://chromiumcodereview.appspot.com/9348055

[epoger, 2012-02-09 12:33:56]

finally out to the trybots... control group is still
https://chromiumcodereview.appspot.com/9348055

[epoger, 2012-02-09 15:54:29]

This deps roll fails with some serious compile errors on Mac... I am going to
work on a fix and then try again.

summary of trybot results:

linux : green
linux_chromeos : 1 new browser_test failure
  LoginPromptBrowserTest.SupplyRedundantAuthsMultiProfile
linux_layout : no new failures
linux_layout_rel : no new failures
linux_shared : green

mac : SERIOUS COMPILE ERRORS, SEE BELOW
mac_layout : SERIOUS COMPILE ERRORS, SEE BELOW
mac_layout_rel : SERIOUS COMPILE ERRORS, SEE BELOW

win : BOTH DEPS ROLL AND CONTROL GROUP FAIL TO LINK REMOTING_SERVICE.EXE
win_layout : STILL RUNNING
win_layout_rel : timed out during build, as usual



Here are the compile errors I saw on all Mac trybots.

In file included from skia/ext/platform_canvas.cc:7:
In file included from ./skia/ext/bitmap_platform_device.h:15:
./skia/ext/bitmap_platform_device_mac.h:64:16: error: virtual function
'onAccessBitmap' has a different return type ('void') than the function it
overrides (which has return type 'const SkBitmap &')
  virtual void onAccessBitmap(SkBitmap*) OVERRIDE;
               ^
./third_party/skia/include/core/SkDevice.h:289:29: note: overridden virtual
function is here
    virtual const SkBitmap& onAccessBitmap(SkBitmap*);
                            ^

/b/build/slave/mac_layout/build/src/skia/ext/bitmap_platform_device_mac.cc:236:39:
error: out-of-line definition of 'skia::BitmapPlatformDevice::onAccessBitmap'
differs from the declaration in the return type
const SkBitmap& BitmapPlatformDevice::onAccessBitmap(SkBitmap* bitmap) {
                                      ^
../skia/ext/bitmap_platform_device_mac.h:64:16: note: previous declaration is
here
  virtual void onAccessBitmap(SkBitmap*) OVERRIDE;
               ^

[caryclark, 2012-02-09 16:06:05]

Looks like the win build may have a similar problem -- or if it is
doesn't cause a compile error, it still may cause a runtime bug:

/chrome/dev/src/skia/ext/bitmap_platform_device_win.cc:250:void
BitmapPlatformDevice::onAccessBitmap(SkBitmap* bitmap) {
/chrome/dev/src/skia/ext/bitmap_platform_device_win.h:65:  virtual
void onAccessBitmap(SkBitmap* bitmap) OVERRIDE;

On Thu, Feb 9, 2012 at 10:54 AM,  <epoger@google.com> wrote:
> This deps roll fails with some serious compile errors on Mac... I am going
> to
> work on a fix and then try again.
>
> summary of trybot results:
>
> linux : green
> linux_chromeos : 1 new browser_test failure
>  LoginPromptBrowserTest.SupplyRedundantAuthsMultiProfile
> linux_layout : no new failures
> linux_layout_rel : no new failures
> linux_shared : green
>
> mac : SERIOUS COMPILE ERRORS, SEE BELOW
> mac_layout : SERIOUS COMPILE ERRORS, SEE BELOW
> mac_layout_rel : SERIOUS COMPILE ERRORS, SEE BELOW
>
> win : BOTH DEPS ROLL AND CONTROL GROUP FAIL TO LINK REMOTING_SERVICE.EXE
> win_layout : STILL RUNNING
> win_layout_rel : timed out during build, as usual
>
>
>
> Here are the compile errors I saw on all Mac trybots.
>
> In file included from skia/ext/platform_canvas.cc:7:
> In file included from ./skia/ext/bitmap_platform_device.h:15:
> ./skia/ext/bitmap_platform_device_mac.h:64:16: error: virtual function
> 'onAccessBitmap' has a different return type ('void') than the function it
> overrides (which has return type 'const SkBitmap &')
>  virtual void onAccessBitmap(SkBitmap*) OVERRIDE;
>               ^
> ./third_party/skia/include/core/SkDevice.h:289:29: note: overridden virtual
> function is here
>    virtual const SkBitmap& onAccessBitmap(SkBitmap*);
>                            ^
>
>
/b/build/slave/mac_layout/build/src/skia/ext/bitmap_platform_device_mac.cc:236:39:
> error: out-of-line definition of
> 'skia::BitmapPlatformDevice::onAccessBitmap'
> differs from the declaration in the return type
> const SkBitmap& BitmapPlatformDevice::onAccessBitmap(SkBitmap* bitmap) {
>                                      ^
> ../skia/ext/bitmap_platform_device_mac.h:64:16: note: previous declaration
> is
> here
>  virtual void onAccessBitmap(SkBitmap*) OVERRIDE;
>
>               ^
>
>
> https://chromiumcodereview.appspot.com/9346023/
>
> --
> You received this message because you are subscribed to the Google Groups
> "skia-dev" group.
> To post to this group, send email to skia-dev@google.com.
> To unsubscribe from this group, send email to
> skia-dev+unsubscribe@google.com.
> For more options, visit this group at
> http://groups.google.com/a/google.com/group/skia-dev/?hl=en.
>

[epoger, 2012-02-09 16:12:06]

On 2012/02/09 16:06:05, caryclark wrote:
> Looks like the win build may have a similar problem -- or if it is
> doesn't cause a compile error, it still may cause a runtime bug:
> 
> /chrome/dev/src/skia/ext/bitmap_platform_device_win.cc:250:void
> BitmapPlatformDevice::onAccessBitmap(SkBitmap* bitmap) {
> /chrome/dev/src/skia/ext/bitmap_platform_device_win.h:65:  virtual
> void onAccessBitmap(SkBitmap* bitmap) OVERRIDE;
> 

Thanks for finding that, Cary!  I will make sure both Mac and Windows look
reasonable before submitting more trybots.

[epoger, 2012-02-09 16:53:50]

out to the trybots again with Mac fix; control group at
https://chromiumcodereview.appspot.com/9348055 going out to trybots again too

[epoger, 2012-02-09 20:18:47]

The trybot results are back and they look fine (summarized below).  Can I get an
LG?  (and then I will commit if the tree reopens soon, otherwise I will wait
until tomorrow morning)

linux : green
linux_chromeos : 1 new failed browser_test, looks like a flake:
  BrowserTest.CancelBeforeUnloadResetsURL
linux_layout : no new failures
linux_layout_rel : no new failures
linux_shared : green

mac : green
mac_layout : 1 new failed webkit_test:
  fast/repaint/fixed-right-in-page-scale.html = IMAGE
mac_layout_rel : no new failures

win : still failing to link remoting_service.exe in both deps roll and control
trybots; I don't understand why we don't see this failure in the waterfall.
win_layout : no new failures
win_layout_rel : timed out in compile step (again…)

[bsalomon, 2012-02-09 20:30:30]

On 2012/02/09 20:18:47, epoger wrote:
> The trybot results are back and they look fine (summarized below).  Can I get
an
> LG?  (and then I will commit if the tree reopens soon, otherwise I will wait
> until tomorrow morning)
> 
> linux : green
> linux_chromeos : 1 new failed browser_test, looks like a flake:
>   BrowserTest.CancelBeforeUnloadResetsURL
> linux_layout : no new failures
> linux_layout_rel : no new failures
> linux_shared : green
> 
> mac : green
> mac_layout : 1 new failed webkit_test:
>   fast/repaint/fixed-right-in-page-scale.html = IMAGE
> mac_layout_rel : no new failures
> 
> win : still failing to link remoting_service.exe in both deps roll and control
> trybots; I don't understand why we don't see this failure in the waterfall.
> win_layout : no new failures
> win_layout_rel : timed out in compile step (again…)

LGTM

[epoger, 2012-02-09 21:08:00]

The tree was just reopened, but I'm going to wait until tomorrow morning.  I
think there are a lot of pent-up changes that will be committed in the next few
minutes, and I want to steer clear of the madness.