Implement generator for beer-song

[kytrinyx]

I've submitted a pull request to update the canonical data for the beer-song exercise (exercism/problem-specifications#661). In order to make it easier to pull in those changes once the PR has been merged, I've created a generator for the beer-song exercise.

In addition to implementing the logic for the generator, I ran the bin/generate beer-song command to create a fresh implementation.

The structure of the canonical data is a bit different from the implementation that the Ruby track had:

Before:

BeerSong.new.verse(n)
BeerSong.new.verses(n, m)
BeerSong.new.lyrics

After:

BeerSong.new.verse(n)
BeerSong.new.verses(n, m)

In the old version, lyrics was a simple indirection, calling verses with the full range.

This is a breaking change, and the book-keeping has been updated to reflect this.

As I implemented the generator, I discovered that the documentation was outdated. This has been updated in #552

[kotp]

This may need to be updated after the exercism/problem-specifications#661 is brought in.

[kytrinyx]

I've updated this per the newest canonical data.

[kotp]

Using the <<-TEXT means that we can have the ending delimiter indented.

At the moment the ending delimiter is on the 1 column, so this is not needed. See comment regarding the generator for note.

[kotp]

Because of the <<-TEXT we could indent the ending delimiter to the level of line 11.

  def <%= test_case.test_name %>
    <%= test_case.skipped %>
    expected = <<-TEXT
<%= test_case.expected %>
    TEXT
    assert_equal expected, <%= test_case.workload %>
  end
<% end %>

The other choice would be to remove the minus before the delimiter and leave the ending delimiter as it is.

[kytrinyx]

Oh, right. Yepp. I don't have strong feelings either way, but I think I'll indent the ending delimiter.

[kotp]

This looks good. We could bring it in now and there would be no foreseen problems.

One small change in the template file is possible, either removing the heredoc delimiter prefix or using that prefix and indenting the ending delimiter.

[kytrinyx]

I've updated the indentation of the closing token for the heredoc.

I'm using the property to decide the name of the function, so as long as that doesn't change over in our canonical data schema discussion, I think we're good here.

[behrtam]

I'm just curious ... wouldn't it make more sense now that the canonical test data has a real version (beer-song/canonical-data.json#L3) to use that instead of the sha-1 hash of the commit (# Common test data version: 9f3d48a)?

[kotp]

I think the merge was done before the canonical was finished... but I think using that canonical version now makes sense to reference.

[kytrinyx]

Yeah, the canonical version makes much more sense. I didn't think of it.