fix: zero value incorect comparison

.github/workflows/ci.yaml

@@ -2,6 +2,12 @@ name: CI
 
 on:
   pull_request:
+    paths:
+      - "**.go"
+      - go.mod
+      - go.sum
+      - ".github/workflows/ci.yaml"
+      - ".golangci.yaml"
 
 jobs:
   lint:

.github/workflows/codecov.yaml

@@ -4,6 +4,11 @@ on:
   push:
     branches:
       - master
+    paths:
+      - "**.go"
+      - go.mod
+      - go.sum
+      - ".github/workflows/codecov.yaml"
 
 jobs:
   test:

README.md

@@ -4,6 +4,7 @@
 [![Go Report Card](https://goreportcard.com/badge/github.com/quagmt/udecimal)](https://goreportcard.com/report/github.com/quagmt/udecimal)
 [![codecov](https://codecov.io/gh/quagmt/udecimal/graph/badge.svg?token=662ET843EZ)](https://codecov.io/gh/quagmt/udecimal)
 [![GoDoc](https://pkg.go.dev/badge/github.com/quagmt/udecimal)](https://pkg.go.dev/github.com/quagmt/udecimal)
+[![Awesome Go](https://awesome.re/mentioned-badge.svg)](https://github.com/avelino/awesome-go#financial)
 
 High performance, high precision, zero allocation fixed-point decimal number for financial applications.
 
@@ -139,7 +140,7 @@ type bint struct {
 	// For coefficients exceeding u128
 	bigInt *big.Int
 
-	// For coefficients less than 2^128-1
+	// For coefficients less than 2^128
 	u128 u128
 }
 ```

decimal.go

@@ -169,6 +169,14 @@ func NewFromHiLo(neg bool, hi uint64, lo uint64, prec uint8) (Decimal, error) {
 
 // newDecimal return the decimal
 func newDecimal(neg bool, coef bint, prec uint8) Decimal {
+	if coef.IsZero() {
+		// make Zero consistent and avoid unexpected cases, such as:
+		// - coef = 0 and neg is true
+		// - coef = 0 and prec != 0
+		// These cases results in incorrect comparison between zero values
+		return Zero
+	}
+
 	return Decimal{neg: neg, coef: coef, prec: prec}
 }
 
@@ -678,12 +686,12 @@ func (d Decimal) rescale(prec uint8) Decimal {
 
 // Neg returns -d
 func (d Decimal) Neg() Decimal {
-	return Decimal{neg: !d.neg, coef: d.coef, prec: d.prec}
+	return newDecimal(!d.neg, d.coef, d.prec)
 }
 
 // Abs returns |d|
 func (d Decimal) Abs() Decimal {
-	return Decimal{neg: false, coef: d.coef, prec: d.prec}
+	return newDecimal(false, d.coef, d.prec)
 }
 
 // Sign returns:

decimal_test.go

@@ -520,9 +520,9 @@ func TestAdd(t *testing.T) {
 		{"9999999999999999999.9999999999999999999", "999999999999999999.999", false},
 		{"9999999999999999999.9999999999999999999", "9999999999999999999.9999999999999999999", false},
 		{"-9999999999999999999.9999999999999999999", "-9999999999999999999.9999999999999999999", false},
+		{"1234567890123456789012345678901234567890.1234567890123456789", "-1234567890123456789012345678901234567890.1234567890123456789", false},
+		{"-1234567890123456789012345678901234567890.1234567890123456789", "1234567890123456789012345678901234567890.1234567890123456789", false},
 		{"1234567890123456789012345678901234567890.1234567890123456789", "1234567890123456789012345678901234567890.1234567890123456789", true},
-		{"1234567890123456789012345678901234567890.1234567890123456789", "-1234567890123456789012345678901234567890.1234567890123456789", true},
-		{"-1234567890123456789012345678901234567890.1234567890123456789", "1234567890123456789012345678901234567890.1234567890123456789", true},
 		{"-1234567890123456789012345678901234567890.1234567890123456789", "-1234567890123456789012345678901234567890.1234567890123456789", true},
 	}
 
@@ -644,10 +644,10 @@ func TestSub(t *testing.T) {
 		{"-9999999999999999999.9999999999999999999", "0.999", false},
 		{"0.999", "-9999999999999999999.9999999999999999999", false},
 		{"-0.999", "9999999999999999999.9999999999999999999", false},
-		{"1234567890123456789012345678901234567890.1234567890123456789", "1234567890123456789012345678901234567890.1234567890123456789", true},
+		{"1234567890123456789012345678901234567890.1234567890123456789", "1234567890123456789012345678901234567890.1234567890123456789", false},
+		{"-1234567890123456789012345678901234567890.1234567890123456789", "-1234567890123456789012345678901234567890.1234567890123456789", false},
 		{"1234567890123456789012345678901234567890.1234567890123456789", "-1234567890123456789012345678901234567890.1234567890123456789", true},
 		{"-1234567890123456789012345678901234567890.1234567890123456789", "1234567890123456789012345678901234567890.1234567890123456789", true},
-		{"-1234567890123456789012345678901234567890.1234567890123456789", "-1234567890123456789012345678901234567890.1234567890123456789", true},
 	}
 
 	for _, tc := range testcases {
@@ -2215,3 +2215,78 @@ func TestInexactFloat64(t *testing.T) {
 		})
 	}
 }
+
+func TestCmpZeroResult(t *testing.T) {
+	testcases := []string{
+		"0",
+		"0.123456789",
+		"-0.123456789",
+		"-123456789.123456789",
+		"1234567890123456789.1234567890123456789",
+		"-1234567890123456789.1234567890123456789",
+		"123456789123456789123456789.1234567890123456789",
+		"-123456789123456789123456789.1234567890123456789",
+	}
+
+	for _, tc := range testcases {
+		t.Run(fmt.Sprintf("cmpZero(%s)", tc), func(t *testing.T) {
+			a := MustParse(tc)
+			cmpZero(t, a, a)
+		})
+	}
+}
+
+func cmpZero(t *testing.T, a, b Decimal) {
+	a1 := a.Sub(b)
+	a2 := a.Add(b.Neg())
+	a3 := Zero.Mul(a)
+	a4 := a.Mul(Zero)
+
+	var (
+		a5  Decimal
+		err error
+	)
+
+	if !a.IsZero() {
+		a5, err = Zero.Div(a)
+		require.NoError(t, err)
+	}
+
+	d := []Decimal{a1, a2, a3, a4, a5}
+
+	for _, dd := range d {
+		require.True(t, dd.IsZero())
+		require.False(t, dd.IsNeg())
+		require.False(t, dd.IsPos())
+
+		require.Equal(t, 0, dd.Cmp(Zero))
+		require.Equal(t, 0, Zero.Cmp(dd))
+		require.Equal(t, Zero, dd)
+	}
+}
+
+func TestCmpWithDiffPrec(t *testing.T) {
+	testcases := []struct {
+		a     int64
+		aprec uint8
+	}{
+		{100, 1},
+		{100, 1},
+		{123456789, 3},
+		{-100, 1},
+		{-123456789, 3},
+		{0, 10},
+		{-0, 13},
+	}
+
+	for _, tc := range testcases {
+		t.Run(fmt.Sprintf("cmpWithDiffPrec(%d, %d)", tc.a, tc.aprec), func(t *testing.T) {
+			a := MustFromInt64(tc.a, tc.aprec)
+
+			for i := tc.aprec; i <= maxPrec; i++ {
+				b := a.rescale(i)
+				cmpZero(t, a, b)
+			}
+		})
+	}
+}