Adding performance tests for HQP that represent importance of optimising HistoryItemsForTerms method

components/omnibox/browser/fake_autocomplete_provider_client.cc

Index: components/omnibox/browser/fake_autocomplete_provider_client.cc
diff --git a/components/omnibox/browser/fake_autocomplete_provider_client.cc b/components/omnibox/browser/fake_autocomplete_provider_client.cc
new file mode 100644
index 0000000000000000000000000000000000000000..4930f958e0cc0d3b012ee63464823fbd4a814a5a
--- /dev/null
+++ b/components/omnibox/browser/fake_autocomplete_provider_client.cc
@@ -0,0 +1,51 @@
+// 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.
+
+#include "components/omnibox/browser/fake_autocomplete_provider_client.h"
+
+#include "base/memory/ptr_util.h"
+#include "components/bookmarks/browser/bookmark_model.h"
+#include "components/bookmarks/test/test_bookmark_client.h"
+#include "components/history/core/browser/history_service.h"
+#include "components/history/core/test/history_service_test_util.h"
+#include "components/omnibox/browser/in_memory_url_index.h"
+#include "components/omnibox/browser/in_memory_url_index_test_util.h"
+
+FakeAutocompleteProviderClient::FakeAutocompleteProviderClient()
+    : pool_owner_(3, "Background Pool") {

[Peter Kasting, 2016/10/24 22:28:28]

Why 3 threads?  Do we need a specific number?  More than 1?

[dyaroshev, 2016/10/25 18:11:34]

This is not my code and I couldn't figure it out fast.
I've just factored it out. Here is the patch that introduced it.
https://codereview.chromium.org/1646893003

[Peter Kasting, 2016/10/25 18:27:56]

This is OK for now, then, but can you loop in rohitrao (the author of that) and
ask why we had this so we can ideally add a comment about it?

[dyaroshev, 2016/10/25 23:06:46]

@rohitrao can you please help out?

[rohitrao (ping after 24h), 2016/10/25 23:12:24]

3 was an arbitrary choice.  Grepping for pool_owner_, tests are evenly split
between 2 and 3.  More than one thread might be useful, so that tasks don't end
up running sequentially, but the exact number of worker threads doesn't matter.

[dyaroshev, 2016/10/26 22:29:12]

Added some explanation

+  bookmark_model_ = bookmarks::TestBookmarkClient::CreateModel();
+  if (history_dir_.CreateUniqueTempDir()) {
+    history_service_ =
+        history::CreateHistoryService(history_dir_.GetPath(), true);
+  }
+
+  in_memory_url_index_.reset(new InMemoryURLIndex(
+      bookmark_model_.get(), history_service_.get(), nullptr,
+      pool_owner_.pool().get(), history_dir_.GetPath(), SchemeSet()));
+  in_memory_url_index_->Init();
+}
+
+FakeAutocompleteProviderClient::~FakeAutocompleteProviderClient() = default;

[Peter Kasting, 2016/10/24 22:28:28]

Nit: I'd write this in the header (instead of here in the .cc) if you don't get
warnings about inlined complex destructors.

[dyaroshev, 2016/10/25 18:11:34]

I do get the warnings.

+
+const AutocompleteSchemeClassifier&
+FakeAutocompleteProviderClient::GetSchemeClassifier() const {
+  return scheme_classifier_;
+}
+
+const SearchTermsData& FakeAutocompleteProviderClient::GetSearchTermsData()
+    const {
+  return search_terms_data_;
+}
+
+history::HistoryService* FakeAutocompleteProviderClient::GetHistoryService() {
+  return history_service_.get();
+}
+
+bookmarks::BookmarkModel* FakeAutocompleteProviderClient::GetBookmarkModel() {
+  return bookmark_model_.get();
+}
+
+InMemoryURLIndex* FakeAutocompleteProviderClient::GetInMemoryURLIndex() {
+  return in_memory_url_index_.get();
+}