~m{a: 1} generates atom instead of string

[KristerV]

thanks for the library. i had my homegrown thing which wasn't half as powerful.

there's a bug when using alt names:

iex> Mix.install([:decimal, {:shorter_maps, "~> 2.0"}])
iex> import ShorterMaps
iex> ~m{a}   
%{"a" => 1}
iex> ~m{a: 1}
%{a: 1}

So in ~m{a} the key will be a string and in ~m{a: 1} the key will be an atom.

[KristerV]

i dug through the code (tests more specifically) and found that the syntax that does work is ~m{"a" => 1}. i'm going to assume this is the best approach and not dig in to "fix" this.

[meyercm]

Hi, sorry to have missed you while this was open and you were struggling.

I'm not 100% clear what your initial goal was: if you want the map %{:a => 1}, then you should just use the standard Elixir atom shorthand notation of %{a: 1}

The goal of shortermaps is when your variable names correspond to the keys of a map; so if you had:

iex> a = 1
...> ~M{a}
%{:a => 1}

# -or- 
...> ~m{a}
%{"a" => 1}

in short, ~M is for atom keys, and ~m is for string keys.

Please feel free to re-open if you think there's an issue to fix here.

[KristerV]

oh no, i did indeed supply a bad example that didn't explain my motive.

i want to pattern match. let's say there's an incoming map like

a = %{"a" => 1, "b" => 2}

i want to extract both keys, but if b is already in use i need to rename it. so i'd like to go

~m{a, b: alt_b} = a

I'm expecting that sigil to return a map with two string keys, but actually alt_b is an atom.

# translated to
%{"a" => a, b: alt_b} = a

does this make more sense?

---

edit: I did at one point figure out by reading the tests that the expected syntax for this is ~m{a, "b" => alt_b}. but this isn't very intuitive and it's undocumented. but if this syntax is on purpose then i guess it's just a documentation issue.