Templatize SkPMFloat to support both 1 and 255 biases.

src/effects/SkColorMatrixFilter.cpp

Index: src/effects/SkColorMatrixFilter.cpp
diff --git a/src/effects/SkColorMatrixFilter.cpp b/src/effects/SkColorMatrixFilter.cpp
index 8dc603042c9e10a2c0fc142533ebaedcd4877cd2..600a3992134f2ec7305da3c9a6a3566d1c523db7 100644
--- a/src/effects/SkColorMatrixFilter.cpp
+++ b/src/effects/SkColorMatrixFilter.cpp
@@ -240,20 +240,20 @@ uint32_t SkColorMatrixFilter::getFlags() const {
 }
 
 static Sk4f premul(const Sk4f& x) {
-    float scale = SkPMFloat(x).a();
-    Sk4f pm = x * SkPMFloat(1, scale, scale, scale);
+    float scale = SkPMFloat<1>(x).a();

[reed1, 2015/08/27 16:34:42]

Hmmm, these use-cases are interesting / confusion.

I have x, which contains 4 floats
I want access to the one that "means" alpha
I must specify the bias, just to extract it as a float? It is a little bit of a
surprise.

[mtklein_C, 2015/08/27 17:07:40]

Yep, pretty much.  They're two distinct types now, though in this case the two
constructors that take Sk4f are mechanically identical, so I can see the
confusion.  Imagine there were a call to this->validate() at the end of the
constructor... the bias defines the type.

You can also just write this directly to get the alpha if it makes more sense to
you:
    float scale = x.kth<SK_A32_SHIFT/8>()

Think this indicates we've coupled together being in SkPMColor order and staying
in range too hard?

+    Sk4f pm = x * SkPMFloat<1>(1, scale, scale, scale);
 
 #ifdef SK_DEBUG
-    SkPMFloat pmf(pm);
+    SkPMFloat<1> pmf(pm);
     SkASSERT(pmf.isValid());
 #endif
 
     return pm;
 }
 
-static Sk4f unpremul(const SkPMFloat& pm) {
+static Sk4f unpremul(const SkPMFloat<1>& pm) {
     float scale = 1 / pm.a(); // candidate for fast/approx invert?
-    return pm * SkPMFloat(1, scale, scale, scale);
+    return pm * SkPMFloat<1>(1, scale, scale, scale);
 }
 
 static Sk4f clamp_0_1(const Sk4f& value) {
@@ -285,7 +285,7 @@ void SkColorMatrixFilter::filterSpan(const SkPMColor src[], int count, SkPMColor
         const Sk4f c4 = Sk4f::Load(fTranspose + 16)*Sk4f(1.0f/255);
 
         // todo: we could cache this in the constructor...
-        SkPMColor matrix_translate_pmcolor = SkPMFloat(premul(clamp_0_1(c4))).round();
+        SkPMColor matrix_translate_pmcolor = SkPMFloat<1>(premul(clamp_0_1(c4))).round();
 
         for (int i = 0; i < count; i++) {
             const SkPMColor src_c = src[i];
@@ -294,7 +294,7 @@ void SkColorMatrixFilter::filterSpan(const SkPMColor src[], int count, SkPMColor
                 continue;
             }
 
-            SkPMFloat srcf(src_c);
+            SkPMFloat<1> srcf(src_c);
 
             if (0xFF != SkGetPackedA32(src_c)) {
                 srcf = unpremul(srcf);
@@ -309,7 +309,7 @@ void SkColorMatrixFilter::filterSpan(const SkPMColor src[], int count, SkPMColor
             Sk4f dst4 = c0 * r4 + c1 * g4 + c2 * b4 + c3 * a4 + c4;
 
             // clamp, re-premul, and write
-            dst[i] = SkPMFloat(premul(clamp_0_1(dst4))).round();
+            dst[i] = SkPMFloat<1>(premul(clamp_0_1(dst4))).round();
         }
     } else {
         const State& state = fState;