Dear Elixir hobbyists and professionals,
I understand that t()
is used for types. For example I can defstruct id: 0
in MyModule
and specify @type t :: %__MODULE__{id: integer)
and then from a different module I can use it as MyModule.t()
as type.
I’m interested to understand what exactly in Elixir/Erlang the t()
is.
In IEX I tried h String.t
, h String.t/0
and i String.t
but it doesn’t work so I’m assuming t()
is not a function even it looks like a function.
If t()
is not a function or atom, could you please tell me what is it?
Is OK to create t()
function or define struct field as t
- defstruct [:t]
? I’m not saying I want to do that, it would be just interesting to understand if I could do that ;-).
Thank you!
Best Regards,
Dan
2 Likes
It’s a way to reference types. It might look like a function call, but everything after @type
, @spec
, … deals with types and types only. It’s not going become executable code.
For a more complete explanation take a look at this one: https://hexdocs.pm/elixir/typespecs.html
1 Like
Thank you LostKobrakai. I read typespecs on hexdocs before I asked. I was just thinking that there is underlying structure. For example String (if you just type it to IEX) is actually an atom, Struct is a map with extra fields…
If there is no underlying structure (like it’s not a function) it means that I can actually create a function t() or struct field “t” without braking something.
1 Like
t()
is basically a convention. You can just as easily have this module:
defmodule MyStruct do
@type my_struct :: %__MODULE__{}
defstruct [:x, :y, :z]
end
And then reference the type in foreign typespecs (= typespecs in different modules) like so:
def ModuleUsingMyStruct do
@spec use_the_struct(MyStruct.my_struct()) :: map()
def use_the_struct(%MyStruct{} = x) do
Map.from_struct(x)
end
end
You can see for yourself that this introduces some annoying extra typing so a convention emerged where t()
is being used, namespaced within the containing module (example from standard lib: String.t()
).
Yes, you can.
5 Likes