doc.go

// Package plenc provides an efficient serialisation protocol based on protobuf,
// but without requiring .proto files or awkward autogenerated structs. Your
// structs are the basis for the serialisation.
//
// plenc needs you to annotate your structs with a plenc tag on each field. The
// tag either indicates that the field should not be encoded, or provides a
// persistent index number that's used for that field in the encoding. The
// indexes within a struct must all be unique and should not be changed. You may
// remove fields, but you should not re-use the index number of a removed field.
// You should not change the type of a field. You can change field names as
// these are not used in the encoding.
//
// Tags look like the following.
//
//  type mystruct struct {
//      A  int `plenc:"1"`
//      B  string `plenc:"-"` // This field is not encoded
//      C  float64 `plenc:"2"`
//      // The values of this field are interned. This reduces allocations if
//      // there are a limited set of distinct values used.
//      D  string `plenc:"3,intern"`
//  }
//
// The plenctag tool will add tags to structs for you.
//
// plenc only encodes fields that are exported - ones where the field name
// begins with a capital letter.
//
// Once your structs have plenc tags, encoding and decoding data is very much
// like the standard JSON library using Marshal and Unmarshal calls. The one
// difference is that the Marshal function allows you to append encoded data to
// an existing slice.
//
package plenc