AcmReceiver: Eliminate AcmReceiver::decoders_

webrtc/modules/audio_coding/neteq/decoder_database.h

Index: webrtc/modules/audio_coding/neteq/decoder_database.h
diff --git a/webrtc/modules/audio_coding/neteq/decoder_database.h b/webrtc/modules/audio_coding/neteq/decoder_database.h
index 3728d1da90c04d68ed52eb09e451eaaef36a29cd..296d059f733d91bd231c7f41854722ffa35bdf7b 100644
--- a/webrtc/modules/audio_coding/neteq/decoder_database.h
+++ b/webrtc/modules/audio_coding/neteq/decoder_database.h
@@ -64,9 +64,8 @@ class DecoderDatabase {
       return decoder ? decoder->SampleRateHz() : cng_decoder_->sample_rate_hz;
     }
 
-    const SdpAudioFormat& GetFormat() const {
-      RTC_DCHECK(audio_format_);
-      return *audio_format_;
+    const SdpAudioFormat* GetFormat() const {
+      return audio_format_ ? &*audio_format_ : nullptr;

[ossu, 2016/09/20 15:30:56]

I don't think we should be able to keep anything in this database that doesn't
have a valid format. My change in: https://codereview.webrtc.org/2355503002/
makes that so, so maybe put this one in after that one?

[kwiberg-webrtc, 2016/09/20 16:30:29]

I agree. We currently do, though, and it isn't a problem for the stuff I change
in this CL (since only external decoders lack SdpAudioFormat, and I don't need
SdpAudioFormat for them).
That's a good change, but I don't think it needs to be a prerequisite for this
CL. You also touch these lines, so whoever lands his CL last will get an
easy-to-fix conflict and can sort it out so that GetFormat returns a const ref
again.

[ossu, 2016/09/21 10:46:10]

Not that it matters enormously, but if so, why don't you just return a reference
to optional here? That's essentially what you're replicating with making it a
pointer. And we've already got the optional just lying around.

[kwiberg-webrtc, 2016/09/21 11:51:57]

Because of the two equivalent return types const Optional<T>& and const T*, the
latter is (1) more well-known, (2) more general, (3) simpler, and (4) shorter.
Yes, returning const Optional<T>& here would be easier to implement and possibly
a tiny bit faster, but I didn't think it was worth it.

[ossu, 2016/09/21 12:00:56]

I'm not sure I agree with you, but I'm also not sure it's worth another pass
through code review for a bit of code it looks like I'll be removing shortly
anyways.

     }
 
     // Returns true if |codec_type| is comfort noise.