Ensure :name is given in Registry.start_link/1

[SpaceEEC]

Issue

Currently when omitting a :name when starting Registry it will be named nil.

Small example:

Interactive Elixir (1.9.0-dev) - press Ctrl+C to exit (type h() ENTER for help)
iex(1)> {:ok, _pid} = Registry.start_link(keys: :unique)
{:ok, #PID<0.102.0>}
iex(2)> {:ok, _pid} = Registry.register(nil, :foo, :bar)
{:ok, #PID<0.103.0>}
iex(3)> [{_pid, val}] = Registry.lookup(nil, :foo)
[{#PID<0.100.0>, :bar}]

This seems to be because this line defaults name to nil:

elixir/lib/elixir/lib/registry.ex

Line 319 in 3680e85

name = Keyword.get(options, :name)

Solution

I added Keyword.has_key?(options, :name) to the check below the mentioned line to ensure that a name actually was given.

elixir/lib/elixir/lib/registry.ex

Line 321 in 3680e85

unless is_atom(name) do

Small example

Interactive Elixir (1.9.0-dev) - press Ctrl+C to exit (type h() ENTER for help)
iex(1)> Registry.start_link(keys: :unique)
** (ArgumentError) expected :name to be given and to be an atom, got: nil
    (elixir) lib/elixir/lib/registry.ex:322: Registry.start_link/1

It would also be possible to default it to a non-atom value, but that would probably make the error message even more misleading.

Notes

• The error message will be expected :name to be given and to be an atom, got: nil then.
  But nil is a valid name! It just was not explicitly given.

Would another unless/2 with a more explicit error message be appropriate?

• I also added a test for this.
  As there are none for the other options this might not be necessary, or misplaced.

[whatyouhide]

We might be better off with something like this:

name =
  case Keyword.fetch(options, :name) do
    {:ok, name} when is_atom(name) -> name
    {:ok, other} -> raise ArgumentError, "expected :name to be an atom, got: #{inspect(other)}"
    :error -> raise ArgumentError, "expected :name option to be present"
  end

...

Wdyt?

[SpaceEEC]

I think this is a good idea. 👍

[whatyouhide]

Can we quickly add a test for the case where :name is present but not an atom? Then we're good to merge!

[whatyouhide]

Thanks @SpaceEEC! 💟