Lesson 07 · Encoding a tag - Build a Protobuf Decoder

Encoding a tag

To write a field you first write its tag, so today you build the inverse of the decoder: pack a field number and wire type back into a single varint. This is the first thing every encoded field emits.

The goal

Encode a field number and wire type into a tag varint.

Start here - the target

TO DO

Scenario: A field number and wire type pack into a tag

Givenfield number 1 and wire type 2 (Len)

WhenAppendTag encodes them

Thenit produces the single byte 0x0A

Andfield 2 wire type 0 gives 0x10, field 3 wire type 5 gives 0x1D, and field 16 wire type 2 gives the two bytes 0x82, 0x01

Background

Encoding a tag is the mirror of decoding one: combine the field number and wire type with field << 3 | wire, then write the result as a varint using the encoder you already have. For small field numbers the tag is a single byte, which is why protobuf rewards putting your most common fields in numbers 1 through 15 - they cost one tag byte instead of two.

Field 16 is the boundary worth pinning: 16 << 3 | 2 is 130, which is larger than 127, so the tag spills to two varint bytes, 0x82 0x01. That is not a special case in your code - it falls straight out of reusing AppendVarint - but it confirms the tag really is just a varint like any other, and that field numbers above 15 carry a real encoding cost.

Make it work

// shift the field number up by 3 and drop the wire type in the low bits
func AppendTag(buf []byte, field int, wire int) []byte {
  tag := uint64(field)<<3 | uint64(wire)
  return AppendVarint(buf, tag)
}

CheckpointDONE

You can encode a tag from a field number and wire type. Commit and stop here.