IO.inspect with invalid width option raises Inspect.Algebra.format exception

[pguillory]

Elixir and Erlang/OTP versions

Calling IO.inspect with the width option set to an invalid value like :infinite (note the typo, not :infinity) in iex raises a confusing exception.

Operating system

macOS Sequoia 15.4.1

Current behavior

iex(1)> IO.inspect(nil, width: :infinite)
** (FunctionClauseError) no function clause matching in Inspect.Algebra.format/2    
    
    The following arguments were given to Inspect.Algebra.format/2:
    
        # 1
        {:doc_group, "nil", :self}
    
        # 2
        :infinite
    
    Attempted function clauses (showing 1 out of 1):
    
        def format(doc, width) when (is_binary(doc) or doc === :doc_nil or doc === :doc_line or is_tuple(doc) and (elem(doc, 0) === :doc_break or elem(doc, 0) === :doc_collapse or elem(doc, 0) === :doc_color or elem(doc, 0) === :doc_cons or elem(doc, 0) === :doc_fits or elem(doc, 0) === :doc_force or elem(doc, 0) === :doc_group or elem(doc, 0) === :doc_nest or elem(doc, 0) === :doc_string or elem(doc, 0) === :doc_limit)) and (width == :infinity or is_integer(width) and width >= 0)
    
    (elixir 1.18.3) lib/inspect/algebra.ex:1010: Inspect.Algebra.format/2
    (elixir 1.18.3) lib/io.ex:490: IO.inspect/3
    iex:1: (file)


### Expected behavior

Expected behavior is that it prints an error like `Invalid width :infinite, valid values are integers or :infinity`.

Actually, even a match error would be fine, if it was easier to read. The important part of this one (`width == :infinity or is_integer(width)`) is buried too deeply.

[josevalim]

Thank you for the report. That’s one of the downsides of dynamic languages, the errors are not necessarily checked at the boundaries.

Given we are adding a type system, this will be naturally solved as that work progresses, so I will go ahead and close this, but you should expect compile time reports sooner than later!