What is the most idiomatic way to construct structs with complex default values?

why not add another function to handle that?

def with_random_bytes(%Email{} = email) do
  Map.put(email, :bytes, random_bytes())
end

Then in your context or anywhere else you need it can always just call it like:

%Email{}
|> Email.with_random_bytes()
|> ...

Email is a string value, not a struct. How would you do it in this case?

There’s no difference, really.

def with_random_bytes(struct) do
  Map.put(struct, :bytes, random_bytes()
end

…

%MyStruct{email: email}
|> MyStruct.with_random_bytes()

However, I think new is the way to go. Don’t overcomplicate things. Document that new structs should be created using the initializer in order to set dynamic defaults.

Are Map.new, Range.new, and MapSet.new non-idiomatic?

100% agree, Session.new(email) or Session.new(opts) with documentation feels right IMO.

I would do it like this to be able to have a default if not passed:

def new(attrs) do
  with %{bytes: nil} = struct <- struct(__MODULE__, attrs) do
    %{struct | bytes: random_bytes()}
  end
end

This way if you pass bytes in it will use whatever bytes you set, however if not set, we can generate it. You can “mimic” the default value.

iex(1)> MyModule.new(%{email: "hello@world.com"})
%MyModule{email: "hello@world.com", bytes: "fjhdsjhkhd23ddw3er"}

iex(2)> MyModule.new(%{email: "hello@world.com", bytes: "123abc"})
%MyModule{email: "hello@world.com", bytes: "123abc"}

Thank you, I like this option. Another question, is it idiomatic to use __MODULE__ when accessing the current module name in order to avoid aliasing the module from within itself?

It certainly is! I do it all the time

I see it as a personal preference thing. I usually alias the module I am in at the top and use that across the module.
In your case I did not know what your module was called, so i just used that for the example.

I like the Modules name mostly because in code reviews it reads easier to me

I just noticed something else I’m not sure of. Are you able to explain the %{bytes: nil} = struct part? It looks like I’m pattern marching against something that does not exist. Is there something special about how the with statement works in this case?

Does it have something to do with the ← in the with statement? Is it similar to what ← does in a comprehension?

Sorry I’m not familiar with “with” statements, watching a video on them now!

Sure no problem,

I pattern match “after” I create the struct because I know the struct has that key always present in atom format. So the with tries to match the struct with a nil for :bytes, if it does match, then i return an updated version with the struct (This is why i do the = struct to be able to grab the struct in order to update/return it in the do block). However, If it does not match, withs without an else automatically return the result; which in our case is the struct