** (FunctionClauseError) no function clause matching in Earmark.Helpers.AstHelpers.augment_tag_with_ial/2

[wojtekmach]

text = "* {:error, {IncompatibleUnitError, message}}"
Earmark.as_ast(text)

results in:

** (FunctionClauseError) no function clause matching in Earmark.Helpers.AstHelpers.augment_tag_with_ial/2

    The following arguments were given to Earmark.Helpers.AstHelpers.augment_tag_with_ial/2:

        # 1
        []

        # 2
        %{}

    Attempted function clauses (showing 1 out of 1):

        def augment_tag_with_ial([{t, a, c} | tags], atts)

    (earmark 1.4.4) lib/earmark/helpers/ast_helpers.ex:12: Earmark.Helpers.AstHelpers.augment_tag_with_ial/2
    (earmark 1.4.4) lib/earmark/ast/inline.ex:214: Earmark.Ast.Inline.converter_for_inline_ial/1
    (elixir 1.10.2) lib/enum.ex:3333: Enum.find_value_list/3
    (earmark 1.4.4) lib/earmark/ast/inline.ex:24: Earmark.Ast.Inline._convert/4
    (earmark 1.4.4) lib/earmark/ast_renderer.ex:153: Earmark.AstRenderer.render_block/3
    (earmark 1.4.4) lib/earmark/ast_renderer.ex:25: Earmark.AstRenderer._render/3
    (earmark 1.4.4) lib/earmark/ast_renderer.ex:143: Earmark.AstRenderer.render_block/3
    (earmark 1.4.4) lib/earmark/ast_renderer.ex:25: Earmark.AstRenderer._render/3

See elixir-lang/ex_doc#1172 for more information.

[RobertDober]

Thank you for analysing and reporting this, I'll try to allocate some time for this.

[wojtekmach]

FWIW:

- * {:error, {IncompatibleUnitError, message}}
+ * `{:error, {IncompatibleUnitError, message}}`

both fixes the bug AND is prettier in the output! But yeah, the bug might be triggered in some other ways so worth fixing.

[RobertDober]

well that is great news as I have no time right now, ok for you to push it into 1.5, meaning, ETA uncertain?

[wojtekmach]

sounds good to me! since it's easy to fix in the user code I don't think it's a big deal.

[dinocosta]

Hi 👋

I'm also encountering this same issue and I'm currently unable to generate docs for a phoenix app. I've managed to trace this to a single doc and here's sample documentation which can trigger this bug:

defmodule Sample do
  @doc """
  Creates a personal hello world message.

  ## Examples

  iex> Sample.hello("Mark", %{name: "Mark", age:
    27, info: %{email: "sample@gmail.com", phone:
      "123456789", address: "Around The Corner" } })
  {:ok, %User{...}}
  """
  def hello do
    :world
  end
end

It seems to be a very specific case because using the same documentation but moving things around slightly makes it work again, for example, the following documentation works:

defmodule Sample do
  @doc """
  Creates a personal hello world message.

  ## Examples

  iex> Sample.hello("Mark", %{name: "Mark", age:
    27, info: %{email: "sample@gmail.com", phone: "123456789", address: "Around The Corner" } })
  {:ok, %User{...}}
  """
  def hello do
    :world
  end
end

But, maintaining the same formatting as the first example and just changing the second element of the tuple in the sample response also works:

defmodule Sample do
  @doc """
  Creates a personal hello world message.

  ## Examples

  iex> Sample.hello("Mark", %{name: "Mark", age:
    27, info: %{email: "sample@gmail.com", phone:
      "123456789", address: "Around The Corner" } })
  {:ok, "Sample String"}
  """
  def hello do
    :world
  end
end

I don't know if this is of much help, the code doesn't even match the docs, but wanted to provide at least another example of how this bug can be reproduced so that if you ever decide to take a look into it you can also experiment this case :)

[RobertDober]

@JoaoFCosta thank you for that detailed report, will turn out helpful for sure

[wojtekmach]

Btw, remember code blocks should be indented with 4 spaces, this way they will be nicely printed in ExDoc, IEx, etc, as well as they won’t trigger the bug :)

…

On 27 May 2020, at 12:48, Robert Dober ***@***.***> wrote:  @JoaoFCosta thank you for that detailed report, will turn out helpful for sure — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[RobertDober]

Pushing into 1.4.6 as #342 is more serious and no easy workaround exists.