[android] Add JobScheduler-based BackgroundTaskScheduler.

components/background_task_scheduler/android/java/src/org/chromium/components/background_task_scheduler/BackgroundTaskSchedulerFactory.java

Index: components/background_task_scheduler/android/java/src/org/chromium/components/background_task_scheduler/BackgroundTaskSchedulerFactory.java
diff --git a/components/background_task_scheduler/android/java/src/org/chromium/components/background_task_scheduler/BackgroundTaskSchedulerFactory.java b/components/background_task_scheduler/android/java/src/org/chromium/components/background_task_scheduler/BackgroundTaskSchedulerFactory.java
new file mode 100644
index 0000000000000000000000000000000000000000..abc96f4e7d12f9c0d6d744fd8dbb95b5dc77a5fc
--- /dev/null
+++ b/components/background_task_scheduler/android/java/src/org/chromium/components/background_task_scheduler/BackgroundTaskSchedulerFactory.java
@@ -0,0 +1,37 @@
+// 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.
+
+package org.chromium.components.background_task_scheduler;
+
+import android.os.Build;
+
+import org.chromium.base.ThreadUtils;
+
+/**
+ * A factory for {@link BackgroundTaskScheduler} that ensures there is only ever a single instance.

[David Trainor- moved to gerrit, 2017/02/24 07:19:49]

Do we really ensure this?  Should we make the task scheduler an Impl and package
protected and just expose an interface?

[Peter Beverloo, 2017/02/24 18:07:09]

The BackgroundTaskScheduler's constructor already is package-protected, so I
think we're good?

[nyquist, 2017/02/24 23:41:11]

I guess since the constructor is package protected like Peter says, maybe we're
good for now? I've also added a comment like you suggested in the other class
referring to the getScheduler() method in this class.

+ */
+public final class BackgroundTaskSchedulerFactory {
+    private static BackgroundTaskScheduler sInstance;
+
+    private static BackgroundTaskSchedulerDelegate getSchedulerDelegate() {

[Peter Beverloo, 2017/02/24 18:07:09]

Wild idea: should we mark this with @TargetApi on M to prevent code from
accidentally assuming it works for M? There may be potential users with poor
test coverage of the start-up path, as that's often non-trivial to test.

[nyquist, 2017/02/24 23:41:12]

Good call. I'll do that for now until we have GcmNetworkManager support!

+        if (Build.VERSION.SDK_INT >= Build.VERSION_CODES.M) {
+            return new BackgroundTaskSchedulerJobService();
+        } else {
+            throw new RuntimeException("Not able to schedule through GCM yet.");
+        }
+    }
+
+    /**
+     * @return the current instance of the {@link BackgroundTaskScheduler}. Creates one if none
+     * exist.
+     */
+    public static BackgroundTaskScheduler getScheduler() {
+        ThreadUtils.assertOnUiThread();
+        if (sInstance == null) sInstance = new BackgroundTaskScheduler(getSchedulerDelegate());
+        return sInstance;
+    }
+
+    // Do not instantiate.
+    private BackgroundTaskSchedulerFactory() {}
+}