Made MixerAudioSource a pure interface.

webrtc/modules/audio_mixer/audio_source_with_mix_status.h

Index: webrtc/modules/audio_mixer/audio_source_with_mix_status.h
diff --git a/webrtc/modules/audio_mixer/audio_source_with_mix_status.h b/webrtc/modules/audio_mixer/audio_source_with_mix_status.h
new file mode 100644
index 0000000000000000000000000000000000000000..8e116de6ade7d9083401d1b23e6f6254b0415075
--- /dev/null
+++ b/webrtc/modules/audio_mixer/audio_source_with_mix_status.h
@@ -0,0 +1,45 @@
+/*
+ *  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_MIXER_AUDIO_SOURCE_WITH_MIX_STATUS_H_
+#define WEBRTC_MODULES_AUDIO_MIXER_AUDIO_SOURCE_WITH_MIX_STATUS_H_
+
+#include "webrtc/modules/audio_mixer/audio_mixer_defines.h"
+
+namespace webrtc {
+
+// A class that holds a mixer participant and its mixing status.
+class AudioSourceWithMixStatus {

[the sun, 2016/10/04 20:36:10]

How about making this class internal to the AudioMixerImpl and naming it
AudioSourceState or something?

[ivoc, 2016/10/04 20:39:29]

This class seems a little complicated to me. How about making is_mixed a public
member and removing all the getters and setters?  I think it may also make sense
to move this class into the audio_mixer_impl.h file, since that's the only place
it's used.

[the sun, 2016/10/04 20:57:41]

I'm thinking about this was_mixed and is_mixed business and can't help but find
it unnecessary. What if you instead just have a gain setting per mixer source
(you need the internal mixer source struct to store this of course). Your mixer
loop figures out the current gain for each source (always 1.0 for "anonymous")
and performs fade in/out while it is mixing to the output buffer (delta =
(gain_current - gain_prev)/128, there's not even a need to know which way it is
ramping), finally latching in the target gain.

[aleloi, 2016/10/05 15:18:19]

That is a nice simplification! How about this plan: I change this class into a
struct and move it in to AudioMixerImpl in this CL. Then I add gain to this
struct, make the required changes mixing behavior (it will be lots of changes
again, I think) and maybe remove anonymous participants in the next CL?

[ivoc, 2016/10/05 18:47:13]

Sounds good.

[the sun, 2016/10/05 19:32:11]

sgtm

+ public:
+  explicit AudioSourceWithMixStatus(MixerAudioSource* audio_source);
+  ~AudioSourceWithMixStatus();
+
+  AudioSourceWithMixStatus(const AudioSourceWithMixStatus&) = default;

[the sun, 2016/10/04 20:36:10]

Uhm, really? Who's owning the MixerAudioSource? I don't really like bare
pointers...

[aleloi, 2016/10/05 15:18:19]

I think you need a copy constructor to put this class in STL containers: vector
copies its contents when it is resized.

+
+  // Returns true if the audio source was mixed this mix iteration.
+  bool IsMixed() const;
+
+  // Returns true if the audio source was mixed previous mix
+  // iteration.
+  bool WasMixed() const;
+
+  // Updates the mixed status.
+  void SetIsMixed(const bool mixed);
+  void ResetMixedStatus();
+  MixerAudioSource* audio_source() const;
+
+ private:
+  MixerAudioSource* audio_source_;

[ivoc, 2016/10/04 20:39:29]

I think this can be MixerAudioSource const *

[aleloi, 2016/10/05 15:18:19]

You are right, thanks!

+  bool is_mixed_ = false;

[the sun, 2016/10/04 20:36:10]

RTC_DISALLOW_IMPLICIT_CONSTRUCTORS

[aleloi, 2016/10/05 15:18:19]

Let's just simplify it to a struct with source, mixed flag and later gain?

[the sun, 2016/10/05 19:32:11]

ok!

+};
+
+}  // namespace webrtc
+
+#endif  // WEBRTC_MODULES_AUDIO_MIXER_AUDIO_SOURCE_WITH_MIX_STATUS_H_