Go: What is the difference between nil and empty slice?

Nil slice — var s []T or var s []T = nil; pointer is nil, length and capacity are 0. Empty slice — s := []T{} or make([]T, 0); has a backing array but no elements.

For most operations they behave identically: len, cap, range, and append all work the same way. The distinction matters for JSON serialization and equality comparisons.

1nilSlice := var ([]int)  // declared inline conceptually
2var nilS []int
3emptyS := []int{}
4madeS := make([]int, 0)
5
6// All have len == 0
7fmt.Println(len(nilS), len(emptyS), len(madeS))  // 0 0 0
8// Only nilS == nil is true
9fmt.Println(nilS == nil)     // true
10fmt.Println(emptyS == nil)   // false

JSON marshaling behavior:

1n, _ := json.Marshal(nilS)     // []byte("null")
2e, _ := json.Marshal(emptyS)   // []byte("[]")
3m, _ := json.Marshal(madeS)    // []byte("[]")

When to use:

• Nil — natural zero state; no allocation needed yet.
• Empty — API responses that must return [] not null; initializing from a known source.

Common mistakes:

• Returning nil from an API handler — clients see null instead of [].
• Checking only len(s) == 0 when nil vs empty matters.
• Thinking append(nilSlice, ...) panics — it doesn't, Go handles it.
• Using slices.Equal to compare a nil and empty slice — they are not equal even though len matches.