Added first layer of the echo canceller 3 functionality

webrtc/modules/audio_processing/aec3/render_transfer_buffer.h

+/*
+ *  Copyright (c) 2016 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef WEBRTC_MODULES_AUDIO_PROCESSING_AEC3_RENDER_TRANSFER_BUFFER_H_
+#define WEBRTC_MODULES_AUDIO_PROCESSING_AEC3_RENDER_TRANSFER_BUFFER_H_
+
+#include <memory>
+#include <vector>
+
+#include "webrtc/base/constructormagic.h"
+#include "webrtc/base/swap_queue.h"
+#include "webrtc/modules/audio_processing/render_queue_item_verifier.h"
+
+namespace webrtc {
+
+// Interface which limits access to the RenderTransferBuffer to write access.
+class RenderTransferBufferWriter {
+ public:
+  virtual ~RenderTransferBufferWriter() = 0;

[aleloi, 2016/12/16 15:04:32]

For some reason, the style guide says that dtors should not be pure virtual:
https://google.github.io/styleguide/cppguide.html#Interfaces

[peah-webrtc, 2016/12/20 10:10:26]

Good find! I have removed this class in the new patch.
Done.

[aleloi, 2016/12/20 15:55:35]

If someone knows why, please enlighten me! Here's my hypothesis: a pure virtual
dtor means that all derived classes must implement it. But that usually means
writing '{}' or '=default' in lots of places where the compiler would implicitly
generate the same thing.

The style guide has no rule for always writing a dtor, but a pure virtual dtor
implies a mandatory requirement for exactly that.

[peah-webrtc, 2016/12/21 23:13:49]

That makes sense, but I don't know if that is the reason. It would be really
interesting to know, though.

+  virtual bool Insert(std::vector<float>* frame) = 0;
+};
+
+class RenderTransferBuffer : public RenderTransferBufferWriter {
+ public:
+  RenderTransferBuffer(size_t num_bands, size_t frame_length);
+  ~RenderTransferBuffer() override;
+  bool Insert(std::vector<float>* frame) override;

[hlundin-webrtc, 2016/12/16 10:04:48]

Comment on the fact that frame will be swapped.

[peah-webrtc, 2016/12/20 10:10:26]

 have removed this class in the new patch.
Done.

+  bool Remove(std::vector<float>* frame);

[hlundin-webrtc, 2016/12/16 10:04:48]

And here.

[peah-webrtc, 2016/12/20 10:10:26]

I have removed this class in the new patch.
Done.

+
+ private:
+  static const size_t kQueueSize = 30;

[hlundin-webrtc, 2016/12/16 10:04:48]

constexpr

[ivoc, 2016/12/19 11:30:22]

Move to .cc file?

[peah-webrtc, 2016/12/20 10:10:26]

I have removed this class in the new patch.
Done.

[peah-webrtc, 2016/12/20 10:10:26]

I have removed this class in the new patch.
Done.

+  SwapQueue<std::vector<float>, RenderQueueItemVerifier<float>> queue_;
+  std::vector<float> frame_;

[aleloi, 2016/12/16 15:04:31]

Will frame_ be used later? It's not used at the moment except during
initialization.

[peah-webrtc, 2016/12/20 10:10:26]

Fully true, that should be removed. I have removed this class in the new patch
though.
Done.

+  RTC_DISALLOW_IMPLICIT_CONSTRUCTORS(RenderTransferBuffer);
+};
+
+}  // namespace webrtc
+
+#endif  // WEBRTC_MODULES_AUDIO_PROCESSING_AEC3_RENDER_TRANSFER_BUFFER_H_