eeff

eeff

Hi, I am learning Elixir and recently reading the Supervisor source code, and came accross start_link/2 and start_link/3:

# https://github.com/elixir-lang/elixir/blob/v1.20.1/lib/elixir/lib/supervisor.ex#L733
def start_link(children, options) when is_list(children) do

# https://github.com/elixir-lang/elixir/blob/v1.20.1/lib/elixir/lib/supervisor.ex#L976
def start_link(module, init_arg, options \\ []) when is_list(options)

As far as I understand, the default argument in def start_link(module, init_arg, options \\ []) should cause another start_link/2 to be defined and should result in ambiguity ? Does guard helps the compiler reduce ambiguity and thus the compiler treats it as a second clause for start_link/2 ? If so, the two clauses does not need to be grouped together ?

I tried the following:

defmodule A do
  def f(a), do: "f(#{inspect(a)})"

  def f(a, b \\ []) when is_list(b), do: "f(#{inspect(a)}, #{inspect(b)})"
end

and result in a warning:

➜  /tmp iex test.ex
Erlang/OTP 27 [erts-15.2] [source] [64-bit] [smp:4:4] [ds:4:4:10] [async-threads:1] [jit:ns]

    warning: this clause for f/1 cannot match because a previous clause at line 2 always matches
    │
  4 │   def f(a, b \\ []) when is_list(b), do: "f(#{inspect(a)}, #{inspect(b)})"
    │       ~
    │
    └─ test.ex:4:7

Interactive Elixir (1.19.0-dev) - press Ctrl+C to exit (type h() ENTER for help)

But if I changed the order of the def’s:

defmodule A do
  def f(a, b \\ []) when is_list(b), do: "f(#{inspect(a)}, #{inspect(b)})"

  def f(a), do: "f(#{inspect(a)})"
end

it results in an error:

➜  /tmp iex test.ex
Erlang/OTP 27 [erts-15.2] [source] [64-bit] [smp:4:4] [ds:4:4:10] [async-threads:1] [jit:ns]

    error: def f/1 conflicts with defaults from f/2
    │
  4 │   def f(a), do: "f(#{inspect(a)})"
    │       ^
    │
    └─ test.ex:4:7: A.f/1

** (CompileError) test.ex: cannot compile module A (errors have been logged)
    test.ex:4: (module)

Can anybody point me to related documentation about this behavior difference ?

Showing Posts 1 to 10

hauleth

hauleth

The difference is that in Supervisor example you have provided there is guard, which limits about of cases it will match.

So the expanded code will look like

def a(x) when is_list(x), do: …

def a(x), do: a(x, :default)

def a(x, y), do: …

So as you can see, there is a situation when 1st clause match.

If you swap order of these functions, then it will mean that there would be no option to call first clause.

eeff

eeff OP

I still don’t understand why switching the order would switch between warning and error.

mudasobwa

mudasobwa

Creator of Cure

Sidenote: Please upgrade the toolchain to Elixir1.20.1 / OTP29.

The compiler explicitly handles \\ operator because Erlang does not have default argument values. It does not allow any redeclaration after \\ has been expanded. Look-behind, though, would have a ton of redundant tracking involved, decreasing performance, that’s why it does not blow up immediately.

The error you observe is raised by first-pass compiler, meeting the \\ default parameters within the function redeclaration.

The warning you observe if you swap clauses is the second-pass compiler, simply warning you about unreachable clause, it has nothing to do with \\ because there is no look-behind.

odelbos

odelbos

You need to do it like this:

defmodule A do
  def f(a, b \\ [])

  def f(a, []), do: "f(#{inspect(a)})"
  def f(a, b) when is_list(b), do: "f(#{inspect(a)}, #{inspect(b)})"
end

IO.puts A.f(5)
# Output --> f(5)

IO.puts A.f(5, [:one, :two])
# Output --> f(5, [:one, :two])
eeff

eeff OP

yes, I know about this but this does not answer my question. thank you anyway!

eeff

eeff OP

It does not allow any redeclaration after \\ has been expanded. Look-behind, though, would have a ton of redundant tracking involved, decreasing performance, that’s why it does not blow up immediately.

this does make more sense to me, I guess there are no public documentation about this behavior.

mudasobwa

mudasobwa

Creator of Cure

Here is the documentation:

If a function with default values has multiple clauses, it is required to create a function head (a function definition without a body) for declaring defaults — Modules and functions — Elixir v1.20.2

Just don’t try to hack the proposed syntax. Supervisor also ditched this cludge, that’s why I suggested to use the latest versions of the toolchain.

odelbos

odelbos

It’s because of the declarative f/1 order.
When we write:

def f(a, b \\ [])

I fact the compiler will expand it in two functions:

def f(a), do: f(a, [])  # <-- This is not a declarative f/1 it's the default-argument of f/2
def f(a, b) do: ...

So in the first case

defmodule A do
  def f(a), do: "f(#{inspect(a)})"     # <--- declarative f/1

  def f(a, b \\ []) when is_list(b), do: ...   # <--- will be expanded by the compiler
end

You will have two f/1 but with the declarative f/1 in first position (top-bottom) will match all clauses. So the compiler emit a warning for the unreachable expanded f/1 that handle the default-argument case of f/2.

In the second case:

defmodule A do
  def f(a, b \\ []) when is_list(b), do ...   # <--- will be expanded by the compiler
  def f(a), do: "f(#{inspect(a)})"     # <--- declarative f/1
end

The declarative f/1 is after the expanded f/1 but from the compiler’s perspective, the expanded f/1 is not a declarative f/1 function it’s the mechanism that handle the default-argument of the f/2 function. So when later the compiler encounter the declarative f/1 which is in conflict with the auto-generated f/1 it emit a compilation error.

mudasobwa

mudasobwa

Creator of Cure

What’s “declarative function” and what does mean “in conflict” in regard to function clauses?

odelbos

odelbos

A declarative function is a function written by the coder in contrast with the auto-generated (or meta-programming) function written by the compiler.

In the first case, there isn’t any conflict with the declarative f/1 (meaning the coder want this function), so when calling f/1 the declarative f/1 will be called. The compiler emit a warning saying “hey I’m doing a trick to handle the default-argument case of f/2 but it will not work” because of the declarative f/1 that will match all clauses about f/1).

In the second case, the compiler is doing his trick to handle the default-argument of f/2, so it add a f/1 but this auto-generated f/1 break the intent of the coder who declared his own f/1. This f/1 will never be called because of the auto-generated f/1, so in this case there is a conflict (breaking coder intent). This is why the compiler emit an error (“I’m doing something that break your intent”):

error: def f/1 conflicts with defaults from f/2

PS:

def f([]), do: ...   # <-- not matching all possible clauses of f/1 (only matching [] clause)
def f(5), do: ...

def f(a), do: ...    # <--- matching all possibles clauses
— All posts loaded —

Where Next? Top

Trending in Questions Top

thiagogsr
** (ArgumentError) expected :max_attempts to be a positive integer, got: {:@, [line: 10, column: 19], [{:max_attempts, [line: 10, column:...
New
RSP87
I’m working on a project that simulates the bumbl example in the programming phoenix book. It acts almost like an email client. We have a...
New
nseaSeb
Hello, I know there is an approach for handling lists that allows for optimized traversal, but I can’t recall the specific method (somet...
New
kpanic
Hi everyone, I am toying with the idea of building a “match maker” for giving personal help to people that wants to start coding. I sta...
New
brecabral
Documentation While reading the Scoped Routes section, I noticed that the documentation currently refers to a problem without explainin...
New
velrest
So my question is quite simple and i have found no conclusive answer on forum, google or AI. Should we use :erlang.float for Integer to ...
New
asweet-confluent
I recently noticed that Elixir’s Logger defaults its primary log level to :debug when no :logger, :level application configuration is pre...
New

Other Trending Topics Top

GenericJam
Edit: 2026 May 15 - This post is archived. Mob is alive!! Main docs: mob v0.7.11 — Documentation A bit of explanation for the slightly c...
New
JesseHerrick
Hey, I’m Jesse and I’m the main contributor behind Dexter, a full-featured, lightning-fast Elixir LSP optimized for large codebases. It s...
New
mudasobwa
I am happy to introduce the very α version of the new programming language compiled to BEAM. Welcome Cure. It has literally three kille...
New
marciok
Hi there! We created Gust: A task orchestrator inspired by Airflow. For those who have never heard about Aiflow, it’s a Python-based wor...
New
mhanberg
Hi everyone! The first release candidate for the Expert language server project is now available! We’ve published a press release detai...
New
jimsynz
Beam Bots (or just BB for short) is a framework for building fault-tolerant robotics applications in Elixir using familiar OTP patterns. ...
New

We're in Beta

About us Mission Statement

Options

Thread Display Mode




Thread Preview

Skip Thread Previews