Use faster clamping for Uint8ClampedArray (#422)

std/assembly/typedarray.ts

@@ -15,6 +15,10 @@ import {
   COMPARATOR
 } from "./internal/sort";
 
+function clampToByte(value: i32): i32 {
+  return ~(value >> 31) & (((255 - value) >> 31) | value); // & 255
+}
+
 export class Int8Array extends TypedArray<i8> {
   @lazy static readonly BYTES_PER_ELEMENT: usize = sizeof<i8>();
 
@@ -112,12 +116,12 @@ export class Uint8ClampedArray extends Uint8Array {
 
   @inline @operator("[]=")
   protected __set(index: i32, value: i32): void {
-    super.__set(index, max(min(value, 255), 0));
+    super.__set(index, clampToByte(value));
   }
 
   @inline @operator("{}=")
   protected __unchecked_set(index: i32, value: i32): void {
-    super.__unchecked_set(index, max(min(value, 255), 0));
+    super.__unchecked_set(index, clampToByte(value));
   }
 
   fill(value: u32, start: i32 = 0, end: i32 = i32.MAX_VALUE): Uint8ClampedArray {

tests/compiler/std/typedarray.optimized.wat

@@ -1814,21 +1814,21 @@
   i32.store8 offset=8
  )
  (func $~lib/typedarray/Uint8ClampedArray#__set (; 25 ;) (type $iii_) (param $0 i32) (param $1 i32) (param $2 i32)
-  (local $3 i32)
   local.get $0
   local.get $1
-  local.get $2
   i32.const 255
   local.get $2
-  i32.const 255
-  i32.lt_s
-  select
-  local.tee $3
-  i32.const 0
-  local.get $3
-  i32.const 0
-  i32.gt_s
-  select
+  i32.sub
+  i32.const 31
+  i32.shr_s
+  local.get $2
+  i32.or
+  local.get $2
+  i32.const 31
+  i32.shr_s
+  i32.const -1
+  i32.xor
+  i32.and
   call $~lib/internal/typedarray/TypedArray<u8>#__set
  )
  (func $~lib/internal/typedarray/TypedArray<u8>#__get (; 26 ;) (type $iii) (param $0 i32) (param $1 i32) (result i32)