fix the numeric truncation problem of SumAs function (#112)

compute.go

@@ -4,11 +4,26 @@ package stl4go
 // returns the result as type R, this is useful when T is too small to hold the result.
 // Complexity: O(len(a)).
 func SumAs[R, T Numeric](a []T) R {
-	var total R
-	for _, v := range a {
-		total += R(v)
+	switch zero := T(0); any(zero).(type) {
+	case int8, int16, int32, int, int64:
+		var total int64
+		for _, v := range a {
+			total += int64(v)
+		}
+		return R(total)
+	case uint8, uint16, uint32, uint, uint64, uintptr:
+		var total uint64
+		for _, v := range a {
+			total += uint64(v)
+		}
+		return R(total)
+	default:
+		var total float64
+		for _, v := range a {
+			total += float64(v)
+		}
+		return R(total)
 	}
-	return total
 }
 
 // Sum summarize all elements in a.

compute_test.go

@@ -1,10 +1,29 @@
 package stl4go
 
-import "testing"
+import (
+	"testing"
+)
 
 func Test_SumAs(t *testing.T) {
-	a := Range[uint8](1, 101)
-	expectEq(t, SumAs[int](a), 5050)
+	t.Run("sum uint8 to int", func(t *testing.T) {
+		a := Range[uint8](1, 101)
+		expectEq(t, SumAs[int](a), 5050)
+	})
+
+	t.Run("sum int to uint8", func(t *testing.T) {
+		a := Range[int](1, 101)
+		expectEq(t, SumAs[uint8](a), uint8(5050%256))
+	})
+
+	t.Run("sum int64 to float64", func(t *testing.T) {
+		a := Range[int64](1, 101)
+		expectEq(t, SumAs[float64](a), 5050.)
+	})
+
+	t.Run("sum float64 to int64", func(t *testing.T) {
+		a := Range[float64](1.1, 101.1)
+		expectEq(t, SumAs[int](a), 101.2*50) // 5060
+	})
 }
 
 func Test_Sum(t *testing.T) {