Reject the permissionState promise with NotSupported Exception (1/3)

Reject the permissionState promise with NotSupported Exception if userVisibleOnly is false (1/3)

This change plumbs onError with an error code. It also provides an
artifical empty onError() method for the purpose of the three sided patch and updates all layout tests to use the userVisibleOnly parameter.

1) This CL
2) https://codereview.chromium.org/1133083002 where the new onError flow is used
3) Cleanup CL where the empty onError method is removed and two new Layout tests are added so the exception is actually tested

BUG=479246
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=195180

[Miguel Garcia, 2015-05-11 11:46:54]

Patchset #1 (id:1) has been deleted

[Miguel Garcia, 2015-05-11 11:59:42]

miguelg@chromium.org changed reviewers:
+ peter@chromium.org

[Miguel Garcia, 2015-05-11 11:59:43]

New implementation of https://codereview.chromium.org/1136513002/ where the
decision on whether to throw the exception is done by the embedder.

[Peter Beverloo, 2015-05-11 12:56:15]

lgtm % comments. Thanks!

https://codereview.chromium.org/1130233003/diff/20001/Source/modules/push_mes...
File Source/modules/push_messaging/PushManager.cpp (right):

https://codereview.chromium.org/1130233003/diff/20001/Source/modules/push_mes...
Source/modules/push_messaging/PushManager.cpp:44: // Backfill userVisible since
it's still used by the chrome layer.
nit: This is a TODO.

https://codereview.chromium.org/1130233003/diff/20001/Source/modules/push_mes...
File Source/modules/push_messaging/PushPermissionStatusCallbacks.h (right):

https://codereview.chromium.org/1130233003/diff/20001/Source/modules/push_mes...
Source/modules/push_messaging/PushPermissionStatusCallbacks.h:36: // FIXME:
Remove this once blinks rolls into chrome
nit: s/FIXME/TODO(miguelg). Elsewhere too, Blink switched over now \o/.

https://codereview.chromium.org/1130233003/diff/20001/public/platform/modules...
File public/platform/modules/push_messaging/WebPushProvider.h (right):

https://codereview.chromium.org/1130233003/diff/20001/public/platform/modules...
public/platform/modules/push_messaging/WebPushProvider.h:24: class
WebPushPermissionStatusCallbacks: public
WebPushPermissionStatusCallbacksOriginal {
nit: space before the colon.

https://codereview.chromium.org/1130233003/diff/20001/public/platform/modules...
public/platform/modules/push_messaging/WebPushProvider.h:31: };
Hmm, I haven't seen this mechanism of dealing with updating types in the API
used before, but I like it :-). It could use a comment above line 26 to clarify
that the argument-less onError() method is the one that has to go, but SGTM
otherwise.

[Miguel Garcia, 2015-05-11 13:22:47]

https://codereview.chromium.org/1130233003/diff/20001/Source/modules/push_mes...
File Source/modules/push_messaging/PushManager.cpp (right):

https://codereview.chromium.org/1130233003/diff/20001/Source/modules/push_mes...
Source/modules/push_messaging/PushManager.cpp:44: // Backfill userVisible since
it's still used by the chrome layer.
On 2015/05/11 12:56:15, Peter Beverloo wrote:
> nit: This is a TODO.

Acknowledged.

https://codereview.chromium.org/1130233003/diff/20001/Source/modules/push_mes...
File Source/modules/push_messaging/PushPermissionStatusCallbacks.h (right):

https://codereview.chromium.org/1130233003/diff/20001/Source/modules/push_mes...
Source/modules/push_messaging/PushPermissionStatusCallbacks.h:36: // FIXME:
Remove this once blinks rolls into chrome
On 2015/05/11 12:56:15, Peter Beverloo wrote:
> nit: s/FIXME/TODO(miguelg). Elsewhere too, Blink switched over now \o/.

Done.

https://codereview.chromium.org/1130233003/diff/20001/public/platform/modules...
File public/platform/modules/push_messaging/WebPushProvider.h (right):

https://codereview.chromium.org/1130233003/diff/20001/public/platform/modules...
public/platform/modules/push_messaging/WebPushProvider.h:24: class
WebPushPermissionStatusCallbacks: public
WebPushPermissionStatusCallbacksOriginal {
On 2015/05/11 12:56:15, Peter Beverloo wrote:
> nit: space before the colon.

Done.

https://codereview.chromium.org/1130233003/diff/20001/public/platform/modules...
public/platform/modules/push_messaging/WebPushProvider.h:31: };
On 2015/05/11 12:56:15, Peter Beverloo wrote:
> Hmm, I haven't seen this mechanism of dealing with updating types in the API
> used before, but I like it :-). It could use a comment above line 26 to
clarify
> that the argument-less onError() method is the one that has to go, but SGTM
> otherwise.

Acknowledged.

[Miguel Garcia, 2015-05-11 13:22:54]

The CQ bit was checked by miguelg@chromium.org

[Miguel Garcia, 2015-05-11 13:22:55]

The patchset sent to the CQ was uploaded after l-g-t-m from peter@chromium.org
Link to the patchset: https://codereview.chromium.org/1130233003/#ps40001
(title: " ")

[commit-bot: I haz the power, 2015-05-11 13:23:09]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1130233003/40001

[commit-bot: I haz the power, 2015-05-11 15:08:05]

Committed patchset #2 (id:40001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=195180