Initial stub version of Async Clipboard API

third_party/WebKit/Source/core/frame/NavigatorClipboard.idl

Index: third_party/WebKit/Source/core/frame/NavigatorClipboard.idl
diff --git a/third_party/WebKit/Source/core/frame/NavigatorClipboard.idl b/third_party/WebKit/Source/core/frame/NavigatorClipboard.idl
new file mode 100644
index 0000000000000000000000000000000000000000..a1ee432b97887c66a1b4cf80bcfe7dc21c06b8c8
--- /dev/null
+++ b/third_party/WebKit/Source/core/frame/NavigatorClipboard.idl
@@ -0,0 +1,11 @@
+// Copyright 2017 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// https://w3c.github.io/clipboard-apis/#navigator-interface
+
+[RuntimeEnabled=AsyncClipboard]
+partial interface Navigator {
+  [CallWith=ScriptState, SecureContext, SameObject]

[foolip, 2017/05/15 15:33:27]

Given that you have SecureContext here, it should be necessary to name the wpt
tests foo.https.html. If the pass anyway, something is strange.

[garykac, 2017/05/17 23:51:15]

Interesting. I was running the tests from the w3c/github repo by running ./serve
and navigating (in my local dev build) to http://localhost:8000/clipboard/.

And they all passed.

I just now ran them using:
  blink/tools/run_layout_tests.sh external/wpt/clipboard
and I see the expected failures ("disallowed" or "undefined").

Renaming the files to .https.html fixes the failures.

Using blink/tools/run_layout_tests.sh also ran all the manual tests. Is there a
clever naming convention to prevent that? Adding -manual doesn't seem to do it
(although it does work when running the github version).

OTOH, the manual tests succeeded in blink when I ran them locally, but I don't
know if that'll be true on the bots (it won't be if they run in parallel).

+  readonly attribute Clipboard clipboard;
+};