Changed AudioEncoder::Encode to take an rtc::Buffer* instead of uint8_t* and a maximum size.

webrtc/modules/audio_coding/codecs/g711/audio_encoder_pcm.cc

Index: webrtc/modules/audio_coding/codecs/g711/audio_encoder_pcm.cc
diff --git a/webrtc/modules/audio_coding/codecs/g711/audio_encoder_pcm.cc b/webrtc/modules/audio_coding/codecs/g711/audio_encoder_pcm.cc
index ff61db8e8d00c07cda36c31a440aac6e833d7eb8..d850675244baf5060defd24d1b5e3f39c0780afe 100644
--- a/webrtc/modules/audio_coding/codecs/g711/audio_encoder_pcm.cc
+++ b/webrtc/modules/audio_coding/codecs/g711/audio_encoder_pcm.cc
@@ -80,8 +80,7 @@ int AudioEncoderPcm::GetTargetBitrate() const {
 AudioEncoder::EncodedInfo AudioEncoderPcm::EncodeInternal(
     uint32_t rtp_timestamp,
     rtc::ArrayView<const int16_t> audio,
-    size_t max_encoded_bytes,
-    uint8_t* encoded) {
+    rtc::Buffer* encoded) {
   if (speech_buffer_.empty()) {
     first_timestamp_in_buffer_ = rtp_timestamp;
   }
@@ -90,12 +89,16 @@ AudioEncoder::EncodedInfo AudioEncoderPcm::EncodeInternal(
     return EncodedInfo();
   }
   RTC_CHECK_EQ(speech_buffer_.size(), full_frame_samples_);
-  RTC_CHECK_GE(max_encoded_bytes, full_frame_samples_);
   EncodedInfo info;
   info.encoded_timestamp = first_timestamp_in_buffer_;
   info.payload_type = payload_type_;
   info.encoded_bytes =
-      EncodeCall(&speech_buffer_[0], full_frame_samples_, encoded);
+      encoded->AppendData(MaxEncodedBytes(),
+                          [&] (rtc::ArrayView<uint8_t> encoded) {
+                            return EncodeCall(&speech_buffer_[0],
+                                              full_frame_samples_,
+                                              encoded.data());

[kwiberg-webrtc, 2016/02/25 00:29:04]

Wouldn't it be better to change EncodeCall to also take a Buffer*, and do the
AppendData in there?

[ossu, 2016/02/25 10:39:51]

I thought about it but decided that EncodeCall is basically just an alias for
the underlying C-style function that performs the actual encoding. Rather than
duplicating the AppendData in each implementation, it made sense to keep it
here.

[kwiberg-webrtc, 2016/02/25 12:24:25]

It will stop making sense when you have to keep the MaxEncodedBytes function
just to support this call, I believe.

[kwiberg-webrtc, 2016/02/27 19:21:46]

You never answered this one.

[ossu, 2016/02/29 08:51:23]

It wasn't a question. Also, I didn't understand it. MaxEncodedBytes is not
required for keeping it this way.

[kwiberg-webrtc, 2016/02/29 11:42:47]

Well, to call AppendData you need *some* encoder-specific function that returns
the maximum number of bytes the encoder will produce. Right now, you can use
MaxEncodedBytes. When you remove MaxEncodedBytes, you'll have to use something
else. Or pass a Buffer* to EncodeCall as I suggested.

There's no harm in leaving it like this for now, though, I suppose.

[ossu, 2016/02/29 11:56:52]

Ah, right! I think that's already in place in the virtual BytesPerSample()
function of AudioEncoderPcm. That's what's used to calculate MaxEncodedBytes()
atm. AudioEncoderPcm (and children) seems set-up to encode full_frame_samples_
at a time, so we should be fine.
 

+                          });
   speech_buffer_.clear();
   return info;
 }