protobuf

# Changelog

## v0.13.0

### Enhancements

  * Do not allow encoding structs as a different struct
  * Support OTP 27

### Bug fixes

  * Add optional keys to `Protobuf.decode/2` typespecs
  * Include `elixirpb` files in Hex package
  * Fix JSON encoding of proto3 optional non-nil defaults

## v0.12.0

### Enhancements

  * Add `include_docs=true` protoc option to skip generating files with
    `@moduledoc false`.
  * Add support for the `google.protobuf.Any` type.
  * Raise `Protobuf.EncodeError` instead of `Protobuf.TypeEncodeError` on
    encode errors.
  * Improve raised error for bad wire types.
  * Raise when encoding/decoding invalid UTF-8 strings.

### Bug fixes

  * Fix typespecs for proto3 optional fields.
  * Fix generated code for map fields outside of packages.
  * Fix JSON encoding map validation.
  * Remove `Protobuf.Wire.Varint` from public API.
  * Fix conflicting PbExtension modules.
  * Raise when trying to JSON encode `google.protobuf.Value`
    values `NaN` or `Infinity`.
  * Support `.` in FieldMask when decoding JSON.

### Deprecations

  * Deprecate the `new(!)/0,1` callbacks, create structs with `%Foo{}`
    or `struct(!)/1,2` instead.

## v0.11.0

### Enhancements

  * Improve formatting in generated code.
  * Fix JSON encoding of field masks.
  * Fix return value of `Protobuf.JSON.RFC3339.encode/2`.
  * Improve error message when encoding oneoff.
  * Add support for proto3 optional field.

### Bug fixes

  * Fix encoding root structs with transform module.
  * Fix a bug with negative durations in JSON.
  * Fix enum default value for structs where module is not available.
  * Support transform module in JSON encoder.

## v0.10.0

### Enhancements

  * Add `Protobuf.get_unknown_varints/1`, returns fields that could not be
    decoded.
  * Support JSON mapping for proto2 schemas.
  * No longer warn when decoding unknown enum values.
  * Validate options in `Protobuf.JSON.encode/2`.
  * Raise `Protobuf.DecodeError` instead of `CaseClauseError` when trying to
    decode an unknown wire type.
  * Raise `Protobuf.DecodeError` instead of `MatchError` when trying to
    decode invalid delimeted types.
  * Include protoc-gen-elixir plugin version in generated files.

### Bug fixes

  * Encode fields in the order they are defined in the schema.
  * Encode the value of oneof fields even when it's the default value.
  * Camelization fixes in generated code using the `module_prefix` and
    `package_prefix` options.
  * Add `nil` to oneof typespec union.
  * Skip calling `Type.new/1` for structs with transform modules.

### Deprecations

  * Warn if `@type t` is defined inside a module that calls `use Protobuf`. To
    fix this warning the files can be regenerated with the latest version of
    protobuf.

### Breaking changes

  * Raise an error if only `@type t` or `defstruct` is defined inside a module
    that calls `use Protobuf`.

## v0.9.0

### Enhancements

Most of the enhancements in this release concern JSON support.

  * JSON support for Google types. The library now supports JSON mapping of
    `Google.Protobuf.*` types such as timestamps, durations, type wrappers, and
    so on. In order to use this correctly, you need to manually generate the
    Protobuf Elixir files from the `.proto` files provided by Google since they
    are not shipped with this library.
  * Accept integer numbers for float fields when decoding JSON.
  * Add `Protobuf.encode_to_iodata/1`.
  * Add the `one_file_per_module` CLI option when using the `protoc-gen-elixir`
    plugin.
  * Use type-aware defaults in struct definitions for Protobuf messages. This
    means that now doing `%MyMessage{}` should be equivalent to
    `MyMessage.new()`.
  * Don't generate the `transformer_module/0` callback in generated Protobuf
    modules if the `transformer_module` CLI option is not used.

### Bug fixes

  * Fix enum aliasing in JSON mapping.
  * Fix encoding of default values when using transformer modules.
  * Raise a `Protobuf.DecodeError` when decoding if fields with number `0` are
    encountered.
  * Fix bugs when decoding big varints, integers, and so on (by casting).
  * Cast enum integers to int32 when they overflow when decoding.
  * Skip encoding for fields if they match their user-defined default in proto2.
  * Fix decoding and encoding of packed repeated fields.
  * Don't pack fields that use the `[packed = false]` option explicitly in
    proto3.

## v0.8.0

### Enhancements

  * [JSON](https://developers.google.com/protocol-buffers/docs/proto3#json)
    encoding and decoding support.
  * [Extensions](https://developers.google.com/protocol-buffers/docs/proto#extensions)
    support.
  * [Enum alias](https://developers.google.com/protocol-buffers/docs/proto3#enum)
    (`allow_alias`) support.
  * Faster and more memory-efficient encoding and decoding algorithms.
  * More accurate typespecs for repeated fields and enums.
  * Add `package_prefix` command-line option to control namespacing on generated
    code.
  * Add `transform_module` command-line option with hooks for custom pre-encode
    and post-decode logic.

### Bug fixes

  * Fix decoding of proto2 payloads containing legacy `group` fields. Instead of
    crashing the decoder, those fields are now skipped.
  * Fix compilation warnings on recent versions of Elixir.
  * Fix `new!` to raise `ArgumentError` when given struct and module don't match.

### Breaking changes

  * Dropped support to OTP 21.1 and lower.

## v0.7.1 (2020-01-07)

### Enhancements

  * Add typespec generation for enum fields.
  * Add `new!/1` function to generated modules with strict field validation.
  * Raise `Protobuf.DecodeError` when unable to decode binary input.
  * Add `--version` command-line flag.

## v0.7.0 (2019-12-19)

## v0.6.3 (2019-08-19)