From ffc22175d30109188d9ab959a35d3a7a288146d2 Mon Sep 17 00:00:00 2001
From: reimar <reimar@b3059339-0415-0410-9bf9-f77b7e298cf2>
Date: Fri, 11 Jul 2008 19:00:49 +0000
Subject: [PATCH] Add a comment on shift vs. division

git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@27259 b3059339-0415-0410-9bf9-f77b7e298cf2
---
 libmpcodecs/ad_msadpcm.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/libmpcodecs/ad_msadpcm.c b/libmpcodecs/ad_msadpcm.c
index 38d35884c8..7e5ea7b071 100644
--- a/libmpcodecs/ad_msadpcm.c
+++ b/libmpcodecs/ad_msadpcm.c
@@ -171,6 +171,9 @@ static int ms_adpcm_decode_block(unsigned short *output, unsigned char *input,
     upper_nibble ^= 1;
     SE_4BIT(snibble);
 
+    // should this really be a division and not a shift?
+    // coefficients were originally scaled by for, which might have
+    // been an optimization for 8-bit CPUs _if_ a shift is correct
     predictor = (
       ((sample1[current_channel] * coeff1[current_channel]) +
        (sample2[current_channel] * coeff2[current_channel])) / 64) +