Add the future storage framework for the BudgetDatabase.

chrome/browser/budget_service/budget.proto

+// Copyright 2016 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.
+
+syntax = "proto2";
+
+package chrome_browser_budget_service;

[Peter Beverloo, 2016/06/29 13:37:27]

package budget_service;

[harkness, 2016/06/30 10:41:57]

I much prefer that, but it's not the pattern I see used in other components of
chrome/browser. Looking again, it's half and half, so I'll go with the shorter
one!

+
+// Chrome requires this.

[Peter Beverloo, 2016/06/29 13:37:27]

What is the added value of this comment?

[harkness, 2016/06/30 10:41:57]

The first proto I looked at didn't have the comment, so I didn't include it,
planning to investigate it later. Of course, when I built later, I'd forgotten
that and it took me a while to see why my build was failing. Looking at other
protos, many of them include the comment, which would have saved me time if it
had been in the original file.

+option optimize_for = LITE_RUNTIME;
+
+// Description of budget chuks.

[Peter Beverloo, 2016/06/29 13:37:27]

Please annotate messages with the next available ID, e.g.:

// Next available id: 3.

[Peter Beverloo, 2016/06/29 13:37:27]

What is the added value of this comment?

[harkness, 2016/06/30 10:41:57]

Done.

[harkness, 2016/06/30 10:41:57]

That was commenting v1, deprecated by v2, but I forgot to remove it. Done now.

+message Budget {
+  // How much budget the origin needs to accumulate before it will be allowed to
+  // perform another budget action.
+  optional double debt = 1;
+
+  // The sequence of budget chunks and their expiration times.
+  repeated BudgetChunk budget = 2;
+}
+
+message BudgetChunk {
+  // The amount of budget remaining in this chunk.
+  optional double budget_amount = 1;
+
+  // The timestamp when the budget expires.

[Peter Beverloo, 2016/06/29 13:37:27]

It would be good to document what format the timestamp is in— why double and not
int64? (e.g. Time::ToInternalValue())

[harkness, 2016/06/30 10:41:57]

Added a comment. I think I'm going to want to change it to int once we're not
using the site engagement timestamps.

+  optional double expiration_timestamp = 2;
+}