A little advice I can give you to analyze @gregvaughn 's code, is to use IO.inspect().

If you run the code as is, it’s not very informative, but if you inspect in between steps, you can get a much better understanding of what’s going on.

iex> input
|> IO.inspect(label: "input")
|> Enum.flat_map(&Enum.to_list/1)
|> IO.inspect(label: "Enum.flat_map")
|> Enum.sort_by(&elem(&1, 1))
|> IO.inspect(label: "Enum.sort_by")
|> Enum.map(&elem(&1, 0))

input: [%{"this is a string" => 24}, %{"en français" => 11}]
Enum.flat_map: [{"this is a string", 24}, {"en français", 11}]
Enum.sort_by: [{"en français", 11}, {"this is a string", 24}]
["en français", "this is a string"]

Yes, IO.inspect is great advice.

In this specific case, I converted the given input data into a more uniform/useful data structure, a list of 2 element tuples {string, weight}. Then I sorted by the 2nd element, then I kept the 1st element.

Thanks you for the fast answer! It works very fine, now I need to dig into it a little bit to understand it fully… so I won’t ask for help next time.