Allow Ranges of Char in macros

[postmodern]

Discussion

Currently it is not possible to define a range of Chars in macros.

Example

{% alpha = ('A'..'Z').to_a + ('a'..'z').to_a %}

{% for char in alpha %}
  puts {{ char }}
{% end %}

Error: range begin must be a NumberLiteral, not CharLiteral

Workaround

One could define the range of Chars as a range of bytes. However, most people would not recognize (0x41..0x5a) or (0x61..7a) as A-Z and a-z, so the readability of the code would be decreased.

[straight-shoota]

I can probably guess on your use case by now, but please provide a convincing practical example why that would be useful.

[postmodern]

  {% ascii   = ('\u{00}'..'\u{255}') %}
  {% alpha   = ('a'..'z').to_a + ('A'..'Z').to_a %}
  {% numeric = ('0'..'9').to_a %}
  
  ...
  
    {% separators = ['_', '-'] %}

    {% for char in (ascii - alpha - numeric - separators) %}
      context "when the given String contains a #{ {{ char }}.inspect } character" do
        let(string) { "Foo#{ {{ char.id }} }Bar" }

        it do
          expect {
            subject.underscore(string)
          }.to raise_error(ArgumentError,"cannot convert string to underscored: #{string.inspect}")
        end
      end
    {% end %}

[straight-shoota]

That's a use case, but sorry I don't think it's a very convincing one.
There is so much logic involved in building the test cases that it can't really serve the testing purpose very reliably. You have to be sure that the macro expressions work correctly as you intend them to do. But they're not validated. They could be completely wrong and run invalid test cases that wouldn't catch bugs in the tested method.