Added two more debug macros for logging scalar values to file.

webrtc/modules/audio_processing/logging/aec_logging.h

Index: webrtc/modules/audio_processing/logging/aec_logging.h
diff --git a/webrtc/modules/audio_processing/logging/aec_logging.h b/webrtc/modules/audio_processing/logging/aec_logging.h
index 3cf9ff89edc34f0bc674404e267c00188d44b247..b062913be28944d098caa378196f5b7d8886945d 100644
--- a/webrtc/modules/audio_processing/logging/aec_logging.h
+++ b/webrtc/modules/audio_processing/logging/aec_logging.h
@@ -43,6 +43,20 @@
     (void) fwrite(data, data_size, 1, file);           \
   } while (0)
 
+// Dumps a raw scalar int32 to file.
+#define RTC_AEC_DEBUG_RAW_WRITE_SCALAR_INT32(file, data)             \
+  do {                                                               \
+    int32_t value_to_store = data;                                   \

[ivoc, 2015/12/01 09:47:52]

Would it be possible/desirable to merge this with the macro below by changing
int32_t to auto?

[peah-webrtc, 2015/12/01 12:21:43]

I think you are right in that it could. My only concern is that we miss the
typecheck if the writing to a certain file happens in several places and may
therefore write it in different formats in different places.

Also, afaics the auto makes it up to the compiler to choose the format, right,
and I guess that it would adapt it to the type of value being written.
For instance in:
int16_t luke=32768;
int16_t skywalker=32768;
RTC_AEC_DEBUG_RAW_WRITE_SCALAR_AUTO(luke*skywalker);
I guess that that would store the value in int32, but it is less straightforward
to see that.

My intended use of this is to read the values into matlab using fread, and then
I need to know the value type, and for me it would be easier to read the logs if
I know whether it either an int32_t or a double.

What do you think? Maybe the auto-scheme could be adapted to allow a simple read
functionality when retrieving the stored data?

+    (void) fwrite(&value_to_store, sizeof(value_to_store), 1, file); \
+  } while (0)
+
+// Dumps a raw scalar double to file.
+#define RTC_AEC_DEBUG_RAW_WRITE_SCALAR_DOUBLE(file, data)            \
+  do {                                                               \
+    double value_to_store = data;                                    \
+    (void) fwrite(&value_to_store, sizeof(value_to_store), 1, file); \
+  } while (0)
+

[kwiberg-webrtc, 2015/11/30 14:55:43]

Drive-by: Please consider using inline functions instead of introducing even
more macros. Something like

static inline void blupp(FILE* file, double data) {
#ifdef BLIPP
  (void) fwrite(&data, sizeof(data), 1, file);
#endif
}

ought to work fine.

 // Opens a raw data file for writing using the specified sample rate.
 #define RTC_AEC_DEBUG_RAW_OPEN(name, instance_counter, file) \
   do {                                                       \
@@ -73,6 +87,14 @@
   do {                                                 \
   } while (0)
 
+#define RTC_AEC_DEBUG_RAW_WRITE_SCALAR_INT32(file, data) \
+  do {                                                   \
+  } while (0)
+
+#define RTC_AEC_DEBUG_RAW_WRITE_SCALAR_DOUBLE(file, data) \
+  do {                                                    \
+  } while (0)
+
 #define RTC_AEC_DEBUG_RAW_OPEN(file, name, instance_counter) \
   do {                                                       \
   } while (0)