CHANGELOG.md

Changelog for Elixir v1.5

Elixir v1.5 brings new features, enhancements and bug fixes to Elixir. It is the first release to leverage features added as part of Erlang/OTP 20. It is also the last release that supports Erlang/OTP 18.

UTF-8 atoms, function names and variables

Elixir v1.5 supports non-quoted atoms and variables to be in UTF-8 when using Erlang/OTP 20+. For example:

test "こんにちは世界" do
  assert :こんにちは世界
end

Or:

saudação = "Bom dia!"

Elixir follows the recommendations in Unicode Annex #31 to make the language more accessible to other languages and communities. Identifiers must start with a letter or underscore, optionally followed by letters, digits, and underscores. Here letter means any UTF-8 letter-character (optionally with a combining mark) and digit means a UTF-8 decimal-digit-character. If you're using ASCII, this does what you'd expect.

Examples of valid variables are:

name    josé   _age   まつもと  _42  адрес

Examples of invalid variables include:

name•   a±2  42

Symbols, such as mathematical notations and emoji, are not allowed identifiers.

For a complete reference on Elixir syntax, see the Syntax Reference. For technical details on Unicode support, see Unicode Syntax.

IEx improvements

IEx got many improvements. The autocompletion system is now capable of autocompleting variables and user imports. New helpers have also been added:

• exports/1 lists all exports (functions and macros) in a given module
• open/1 opens up the source of a module or function directly in your editor. For example, open MyApp.Module
• runtime_info/0 prints general information about the running system, such as number of cores, runtime version, allocation of memory in the VM and more

IEx also features a breakpoint system for code debugging. The following functions have been added to aid debugging:

• break!/2 - sets up a breakpoint for a given Mod.fun/arity
• break!/4 - sets up a breakpoint for the given module, function, arity
• breaks/0 - prints all breakpoints and their ids
• continue/0 - continues until the next breakpoint in the same process
• open/0 - opens editor on the current breakpoint
• remove_breaks/0 - removes all breakpoints in all modules
• remove_breaks/1 - removes all breakpoints in a given module
• reset_break/1 - sets the number of stops on the given id to zero
• reset_break/3 - sets the number of stops on the given module, function, arity to zer
• respawn/0 - starts a new shell (breakpoints will ask for permission once more)
• whereami/1 - shows the current location

Exception.blame

Exception.blame/3 is a new function in Elixir that is capable of attaching debug information to certain exceptions. Currently this is used to augment FunctionClauseErrors with a summary of all clauses and which parts of clause match and which ones didn't. For example:

iex> Access.fetch(:foo, :bar)
** (FunctionClauseError) no function clause matching in Access.fetch/2

The following arguments were given to Access.fetch/2:

    # 1
    :foo

    # 2
    :bar

Attempted function clauses (showing 5 out of 5):

    def fetch(-%struct{} = container-, key)
    def fetch(map, key) when -is_map(map)-
    def fetch(list, key) when -is_list(list)- and is_atom(key)
    def fetch(list, key) when -is_list(list)-
    def fetch(-nil-, _key)

(elixir) lib/access.ex:261: Access.fetch/2

In the example above, an argument that did not match or guard that did not evaluate to true are shown between -. If the terminal supports ANSI coloring, they are wrapped in red instead of the - character.

Since blaming an exception can be expensive, Exception.blame/3 must be used exclusively in debugging situations. It is not advised to apply it to production components such as a Logger. This feature has been integrated into the compiler, the command line, ExUnit and IEx.

This feature also requires Erlang/OTP 20+.

Streamlined child specs

Elixir v1.5 streamlines how supervisors are defined and used in Elixir. Elixir now allows child specifications, which specify how a child process is supervised, to be defined in modules. In previous versions, a project using Phoenix would write:

import Supervisor.Spec

children = [
  supervisor(MyApp.Repo, []),
  supervisor(MyApp.Endpoint, [])
]

Supervisor.start_link(children, strategy: :one_for_one)

In Elixir v1.5, one might do:

children = [
  MyApp.Repo,
  MyApp.Endpoint
]

Supervisor.start_link(children, strategy: :one_for_one)

The above works by calling the child_spec/1 function on the given modules.

This new approach allows MyApp.Repo and MyApp.Endpoint to control how they run under a supervisor. This reduces the chances of mistakes being made, such as starting an Ecto repository as a worker or forgetting to declare that tasks are temporary in a supervision tree.

If it is necessary to configure any of the children, such can be done by passing a tuple instead of an atom:

children = [
  {MyApp.Repo, url: "ecto://localhost:4567/my_dev"},
  MyApp.Endpoint
]

The modules Agent, Registry, Task, and Task.Supervisor have been updated to include a child_spec/1 function, allowing them to be used directly in a supervision tree similar to the examples above. use Agent, use GenServer, use Supervisor, and use Task have also been updated to automatically define an overridable child_spec/1 function.

Finally, child specifications are now provided as maps (data-structures) instead of the previous Supervisor.Spec.worker/3 and Supervisor.Spec.supervisor/3 APIs. This behaviour also aligns with how supervisors are configured in Erlang/OTP 18+. See the updated Supervisor docs for more information, as well as the new Supervisor.init/2 and Supervisor.child_spec/2 functions.

@impl

This release also allows developers to mark which functions in a given module are an implementation of a callback. For example, when using the Plug project, one needs to implement both init/1 and call/2 when writing a Plug:

defmodule MyApp do
  @behaviour Plug

  def init(_opts) do
    opts
  end

  def call(conn, _opts) do
    Plug.Conn.send_resp(conn, 200, "hello world")
  end
end

The problem with the approach above is that, once more and more functions are added to the MyApp module, it becomes increasingly harder to know the purposes of the init/1 and call/2 functions. For example, for a developer unfamiliar with Plug, are those functions part of the MyApp API or are they implementations of a given callback?

Elixir v1.5 introduces the @impl attribute, which allows us to mark that certain functions are implementation of callbacks:

defmodule MyApp do
  @behaviour Plug

  @impl true
  def init(_opts) do
    opts
  end

  @impl true
  def call(conn, _opts) do
    Plug.Conn.send_resp(conn, 200, "hello world")
  end
end

You may even use @impl Plug if you want to explicitly document which behaviour defines the callback you are implementing.

Overall, using @impl has the following advantages:

• Readability of the code is increased, as it is now clear which functions are part of your API and which ones are callback implementations. To reinforce this idea, @impl true automatically marks the function as @doc false, disabling documentation unless @doc is explicitly set

• If you define @impl before a function that is not a callback, Elixir will error. This is useful in case of typos or in case the behaviour definition changes (such as a new major version of a library you depend on is released)

• If you use @impl in one implementation, Elixir will force you to declare @impl for all other implementations in the same module, keeping your modules consistent

Calendar improvements

This release brings further improvements to Calendar types. It adds arithmetic and others functions to Time, Date, NaiveDateTime and Datetime as well as conversion between different calendars.

v1.5.3 (2017-12-19)

1. Bug fixes

Elixir

• [Calendar] Consider microseconds in maximum possible Calendar.ISO datetime
• [Enum] Fix chunk_every/4 when step > count
• [Kernel] Warn duplicate definitions in macros
• [Kernel] Remove dialyzer warnings from else in with clauses
• [Kernel] Do not warning on upcoming @deprecated and @since attributes for v1.6
• [MapSet] Return valid MapSet when union-ing a legacy MapSet

ExUnit

• [ExUnit] Fix compiler warnings in assert_receive/3

v1.5.2 (2017-09-29)

1. Enhacements

Elixir

• [Kernel] Optimize function definition with multiple clauses by not traversing the internal clauses table
• [Kernel] Warn if unary operators are followed by new lines
• [Registry] Use the name of the Registry as its :id in the child_spec/1 function

2. Bug fixes

Elixir

• [DateTime] Fix negative microsecond result when passing negative Unix epochs to from_unix/2
• [Kernel] Improve error message for oversized atoms
• [Kernel] Ensure @impl attribute also propagates to clauses from default arguments
• [Kernel] Emit proper error for unknown vars inside binary pattern in match

IEx

• [IEx] Do not crash IEx unexpectedly on System.stop/0
• [IEx.Helpers] Ensure exiting a breakpoint set inside a breakpoint does not terminate the shell unexpectedly

Mix

• [mix local.hex] Ensure --if-missing flag works as advertised
• [mix test] Do not trigger additional error reports when there is a failure when loading test files
• [Mix.SCM.Git] Ensure errors when invoking git propagate correctly

v1.5.1 (2017-08-01)

1. Enhancements

EEx

• [EEx.Engine] Add handle_begin and handle_end to EEx

Elixir

• [Kernel] Do not use references on function/macro definitions - this provides large improvements in compilation times in some rare corner cases
• [Supervisor] Support mixing old and new typespecs in Supervisor.init/2 and Supevisor.start_link/2

Mix

• [mix profile] Allow profile tasks to run without a project

2. Bug fixes

EEx

• [EEx.Engine] Do not re-use the value of the init/1 callback throughout the compilation stack

Elixir

• [Kernel] Ensure dialyzer does not emit warnings in some uses of with
• [Kernel] Fix dialyzer warnings when defmacrop is used in modules
• [Kernel] Ensure Elixir modules can be dialyzed without starting the Elixir application
• [Kernel] Do not serialize references in quoted expressions
• [Kernel] Make sure structs expansion use the latest definition available when struct modules are recompiled
• [Task] Support :infinity timeout on Task streams
• [Typespec] Ensure typespecs allow tuple to be used as variable names

v1.5.0 (2017-07-25)

1. Enhancements

Elixir

• [Access] Optimize Access.get/2
• [Base] Optimise Base encode/decode
• [Calendar] Implement Inspect for DateTime with Calendar.ISO
• [Calendar] Add "ISO days" format for conversions between Calendars and Date.convert/2, Time.convert/2, NaiveDateTime.convert/2 and DateTime.convert/2 (as well as bang variants)
• [Calendar] Add :calendar field to Time struct
• [Calendar] Add Time.diff/3, Date.add/2, Date.diff/2, DateTime.diff/3
• [Calendar] Add Date.range/2
• [Calendar] Add Date.new/4, DateTime.utc_now/1, NaiveDateTime.new/8 and Time.new/5 that allow specifing calendar
• [Enum] Add Enum.chunk_by/4 and Stream.chunk_by/4
• [Enum] Add Enum.chunk_every/2 and Enum.chunk_every/4 with a more explicit API than Enum.chunk/2 and Enum.chunk/4
• [Exception] Add Exception.blame/3 that adds metadata to exceptions
• [File] Add File.read_link/1 and File.read_link!/1
• [File] Introduce :trim_bom option for File.stream!/2
• [Inspect] Add :printable_limit to control the limit of printable structures
• [Integer] Add Integer.gcd/2
• [Kernel] Add left not in right to check that the left side is not in the enumerable on the right
• [Kernel] Use the new debug_info chunk in OTP 20. This provides a mechanism for tools to retrieve the Elixir AST from beam files
• [Kernel] defoverridable/1 accepts a module name as argument and marks all callbacks as overridable
• [Kernel] Allow non-quoted Unicode atoms and variables according to Unicode Annex #31 (see Unicode Syntax document)
• [Kernel] Warn when a :__struct__ key is used when building/updating structs
• [Kernel] Cache the AST on definitions. This speeds up the compilation time from 10% to 15% measured across different projects
• [Kernel] Improve compiler error message on invalid patterns and guards
• [Keyword] Add replace/3 and replace!/3 for replacing an existing key
• [List] List.starts_with?/2
• [Macro] Introduce Macro.generate_arguments/2
• [Map] Optimize Map.merge/3 by choosing merging direction
• [Map] Add replace/3 and replace!/3 for replacing an existing key
• [Map] Raise BadMapError in Map.equal?/2 when either of the two arguments is not a map
• [MapSet] Reduce MapSet size when serialized to approximately half
• [Process] Add Process.cancel_timer/2
• [Protocol] Show available implementations on Protocol.UndefinedError if the protocol has been consolidated
• [Registry] Support ETS guard conditions in Registry.match/3
• [Registry] Support parallel: true in Registry.dispatch/3
• [Registry] Introduce Registry.unregister_match/4
• [Stream] Add Stream.chunk_every/2 and Stream.chunk_every/4 with a more explicit API than Stream.chunk/2 and Stream.chunk/4
• [String] Optimise binary pattern matching in String.split/1 and String.trim_*/1
• [Supervisor] Add Supervisor.init/2 and Supervisor.child_spec/2
• [Supervisor] Allow module and {module, arg} to be given to Supervisor.start_link/2 and invoke module.child_spec(arg) on each argument
• [Task] Support :on_timeout in Task.async_stream to control how tasks are terminated
• [Task] Add ordered: false support to Task.async_stream

ExUnit

• [ExUnit] Show code snippet from test source file in case of test errors
• [ExUnit] Use Exception.blame/3 when formatting test errors
• [ExUnit] Make assert_raise/2 fail if the underlying exception has a broken message/1 implementation
• [ExUnit] Add start_supervised/2 and stop_supervised/1 to ExUnit. Processes started by this function are automatically shut down when the test exits

IEx

• [IEx.Autocomplete] Support autocompletion of variable names
• [IEx.Autocomplete] Support autocompletion of functions imported using import Mod, only: [...]
• [IEx.Evaluator] Use Exception.blame/3 when showing errors in the terminal
• [IEx.Helpers] Add exports/1 IEx helper to list all exports in a module
• [IEx.Helpers] Add break!/2, break!/4, breaks/0, continue/0, open/0, remove_breaks/0, remove_breaks/1, reset_break/1, reset_break/3 and whereami/1 for code debugging
• [IEx.Helpers] No longer emit warnings for IEx commands without parentheses
• [IEx.Helpers] Add runtime_info/0 for printing runtime system information
• [IEx.Helpers] Add open/1 to open the source of a given module/function in your editor
• [IEx.Info] Implement IEx.Info protocol for calendar types

Logger

• [Logger] Add metadata: :all configuration to log all metadata

Mix

• [mix compile.elixir] Add --all-warnings option to Elixir compiler that shows all warnings from the previous compilation (instead of just of the files being compiled)
• [mix escript.build] Strip debug information from escripts by default and add option :strip_beam which defaults to true
• [mix loadpaths] Ensure --no-deps-check do not trigger SCM callbacks (such as git)
• [mix local.hex] Add --if-missing flag to local.hex mix task
• [mix profile.cprof] Add Mix.Tasks.Profile.Cprof for count-based profiling
• [mix new] New styling for generated applications

2. Bug fixes

Elixir

• [Calendar] Ensure Calendar.ISO raises a readable error when reaching up the year 10000 restriction
• [Calendar] Return {:error, :invalid_time} for wrong precision instead of crashing when parsing ISO dates
• [Enumerable] Raise Protocol.UndefinedError on bad functions in Enumerable implementation
• [File] Ensure recursive file operations raise on paths with null bytes (security issue reported by Griffin Byatt)
• [File] Support :ram/:raw files in File.copy/2
• [Inspect] Do not use colors when inspecting error messages
• [Kernel] Support guards on anonymous functions of zero arity
• [Kernel] Fix compilation of maps used as maps keys inside matches
• [Kernel] Ensure do clause in with is tail call optimizable
• [Module] on_definition/6 callback receives body wrapped in a keyword list, such as [do: body]. This solves a bug where it was impossible to distinguish between a bodyless clause and a function that returns nil.
• [Path] Ensure recursive path operations raise on paths with null bytes (security issue reported by Griffin Byatt)
• [Protocol] Do not lose source compile info on protocol consolidation
• [Record] Properly escape quoted expressions passed to defrecord
• [Regex] Fix inspect/2 for regexes with / terminator in them
• [Registry] Ensure Registry.match/4 works with :_ as key
• [Stream] Fix stream cycle over empty enumerable
• [String] Consider Unicode non-characters valid according to the specification in String.valid?/1
• [StringIO] Fix encoding and performance issues in StringIO.get_until
• [System] Raise on paths with null bytes in System.cmd/2 and in System.find_executable/1 (security issue reported by Griffin Byatt)
• [System] Raise on ill-formed environment variables (security issue reported by Griffin Byatt)

ExUnit

• [ExUnit] Properly account failed tests when setup_all fails
• [ExUnit] Having two or more describe blocks with the same name will now raise an error.

IEx

• [IEx] Skip autocompletion of module names that are invalid without being quoted
• [IEx] Skip autocompletion of functions with default arguments with @doc false
• [IEx] Do not start oldshell alongside IEx

Mix

• [mix compile.elixir] Store multiple sources in case of module conflicts. This solves an issue where _build would get corrupted when compiling Elixir projects with module conflicts
• [mix compile.erlang] Do not silently discard Erlang compile errors
• [mix compile.erlang] Properly track -compile module attribute when specified as a list
• [mix compile.protocols] Ensure protocol implementations do not "disappear" when switching between applications in umbrella projects by having separate consolidation paths per project
• [mix compile.protocols] Do not raise when consolidating a protocol that was converted into a module

3. Soft deprecations (no warnings emitted)

Elixir

• [Kernel] not left in right is soft-deprecated in favor of left not in right

4. Deprecations

Elixir

• Atom.to_char_list/1, Float.to_char_list/1, Integer.to_char_list/1, Integer.to_char_list/2, Kernel.to_char_list/1, List.Chars.to_char_list/1, String.to_char_list/1 have been deprecated in favor of their to_charlist version. This aligns with the naming conventions in both Erlang and Elixir
• [Enum] Deprecate Enum.filter_map/3 in favor of Enum.filter/2 + Enum.map/2 or for-comprehensions
• [GenEvent] Deprecate GenEvent and provide alternatives in its docs
• [Kernel] Using () to mean nil is deprecated
• [Kernel] :as_char_lists value in Inspect.Opts.t/0 type, in favor of :as_charlists
• [Kernel] :char_lists key in Inspect.Opts.t/0 type, in favor of :charlists
• [Module] Using Erlang parse transforms via @compile {:parse_transform, _} is deprecated
• [Stream] Deprecate Stream.filter_map/3 in favor of Stream.filter/2 + Stream.map/2
• [String] String.ljust/3 and String.rjust/3 are deprecated in favor of String.pad_leading/3 and String.pad_trailing/3 with a binary padding
• [String] String.strip/1 and String.strip/2 are deprecated in favor of String.trim/1 and String.trim/2
• [String] String.lstrip/1 and String.rstrip/1 are deprecated in favor of String.trim_leading/1 and String.trim_trailing/1
• [String] String.lstrip/2 and String.rstrip/2 are deprecated in favor of String.trim_leading/2 and String.trim_trailing/2 with a binary as second argument
• [Typespec] char_list/0 type is deprecated in favor of charlist/0

EEx

• [EEx] Deprecate <%= in "middle" and "end" expressions, e.g.: <%= else %> and <%= end %>

v1.4

The CHANGELOG for v1.4 releases can be found in the v1.4 branch.