Lesson 12 · Fixed32, little-endian - Build a Protobuf Decoder

Fixed32, little-endian

Some types skip the varint entirely and store exactly four little-endian bytes: fixed32, sfixed32, and float. Today you decode and encode those four bytes, the I32 wire type, and confirm the byte order both ways.

The goal

Decode and encode a fixed32 value as four little-endian bytes.

Start here - the target

TO DO

Scenario: A fixed32 is four little-endian bytes

Giventhe bytes 0xD2, 0x02, 0x96, 0x49

WhenReadFixed32 decodes them as little-endian

Thenthe uint32 value is 1234567890 (0x499602D2)

Andencoding 1234567890 back gives the same four bytes, and the float32 value 1.0 is the bytes 0x00, 0x00, 0x80, 0x3F

Background

The I32 wire type is dead simple: exactly four bytes, no length prefix, stored little-endian (least-significant byte first). It backs fixed32, sfixed32, and float. Decoding is just reassembling the four bytes with the first byte in the low position; encoding writes them back low byte first. Pin the example 0xD2 0x02 0x96 0x49, which is 1234567890 - reversing the bytes gives the hex 0x499602D2, so getting the order wrong is immediately visible.

A float uses the same four bytes, reinterpreted as an IEEE 754 single-precision number rather than an integer - most languages expose this as a bit-cast (Go’s math.Float32frombits, for instance). The value 1.0 has the bit pattern 0x3F800000, which little-endian is 0x00 0x00 0x80 0x3F, and a negative like -2.5 is 0x00 0x00 0x20 0xC0. The integer and the float share the wire; only the schema says which one you meant.

Make it work

// low byte first
func ReadFixed32(b []byte) uint32 {
  return uint32(b[0]) | uint32(b[1])<<8 | uint32(b[2])<<16 | uint32(b[3])<<24
}
func AppendFixed32(buf []byte, v uint32) []byte {
  return append(buf, byte(v), byte(v>>8), byte(v>>16), byte(v>>24))
}
// a float32 is these same 4 bytes reinterpreted (math.Float32frombits)

CheckpointDONE

You can decode and encode fixed32 values. Commit and stop here.