Generate Signed Distance Field directly from vector path

src/core/SkDistanceFieldGen.cpp

Index: src/core/SkDistanceFieldGen.cpp
diff --git a/src/core/SkDistanceFieldGen.cpp b/src/core/SkDistanceFieldGen.cpp
index 147aefad79052428453494e33fb98ea4e21e9d43..49d3b9fd4efe8bc700ad48c97040645f12ad508c 100755
--- a/src/core/SkDistanceFieldGen.cpp
+++ b/src/core/SkDistanceFieldGen.cpp
@@ -317,7 +317,7 @@ static void B2(DFData* curr, int width) {
 
 #if !DUMP_EDGE
 static unsigned char pack_distance_field_val(float dist, float distanceMagnitude) {
-    if (dist <= -distanceMagnitude) {
+    if (dist <= -(distanceMagnitude*(1.0f-1.0f/128.0f))) {

[bsalomon, 2016/02/02 18:03:48]

Can you add a comment that explains this constant?

[Joel.Liang, 2016/02/04 12:12:04]

Done. And I'd like to give a example here:

If dist is -3.99999, the result became 0.
The correct result should be 255.

input:
  distanceMagnitude = 4.0
  dist = -3.99999
result:
  (unsigned char)((distanceMagnitude-dist)*128.0/distanceMagnitude)
  => (unsigned char)((4.0-(-3.99999))*128.0/4.0)
  => (unsigned char)(255.99968)
  => 0

EDIT: I have changed the way we implement this method to more clear about the
constant. The result should be the same as the original one.

[jvanverth1, 2016/02/04 14:21:22]

Thanks, that equation is a lot clearer (and thanks for finding that bug).

[Joel.Liang, 2016/02/22 07:34:00]

Actually, the example I gave here is wrong(The input "dist" not close enough to
-4.0).
But the bug do existed.
The problem was if we pass a value which very close to -4.0 (e.g. -3.9999997f).
And then the result in float is close to 256.0 (e.g. 255.999999)
But the float type in C could not represent this much significant digits.
So the result became 256.0f and then unsigned char 0.

         return 255;
     } else if (dist > distanceMagnitude) {
         return 0;