avfilter/af_dynaudnorm: use M_PI
authorGanesh Ajjanagadde <gajjanagadde@gmail.com>
Wed, 11 Nov 2015 03:26:25 +0000 (22:26 -0500)
committerGanesh Ajjanagadde <gajjanagadde@gmail.com>
Fri, 20 Nov 2015 13:29:56 +0000 (08:29 -0500)
The ad-hoc pi constant has a ludicrous number of digits that offer no
value whatsoever. M_PI is more consistent and readable across the
codebase.

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
libavfilter/af_dynaudnorm.c

index 8f0c2d0..62a2653 100644 (file)
@@ -227,8 +227,6 @@ static int cqueue_pop(cqueue *q)
     return 0;
 }
 
-static const double s_pi = 3.1415926535897932384626433832795028841971693993751058209749445923078164062862089986280348253421170679;
-
 static void init_gaussian_filter(DynamicAudioNormalizerContext *s)
 {
     double total_weight = 0.0;
@@ -238,7 +236,7 @@ static void init_gaussian_filter(DynamicAudioNormalizerContext *s)
 
     // Pre-compute constants
     const int offset = s->filter_size / 2;
-    const double c1 = 1.0 / (sigma * sqrt(2.0 * s_pi));
+    const double c1 = 1.0 / (sigma * sqrt(2.0 * M_PI));
     const double c2 = 2.0 * pow(sigma, 2.0);
 
     // Compute weights