Reduce or remove errors and warnings when parsing invalid UTF-8

[bquorning]

When running against this bit of code:

RSpec.describe "invalid UTF8" do
  let(:invalid_utf8) { "hello \255" }

  it "is not the answer" do
    expect(invalid_utf8).to_not eq(42)
  end
end

I get these errors and warnings:

utf8_spec.rb:4:24: error: literal contains escape sequences incompatible with UTF-8
utf8_spec.rb:4:   let(:invalid_utf8) { "hello \255" }
utf8_spec.rb:4:                        ^~~~~~~~~~~~
Warning! Skipping utf8_spec.rb due to parsing error!
utf8_spec.rb:4:24: error: literal contains escape sequences incompatible with UTF-8
utf8_spec.rb:4:   let(:invalid_utf8) { "hello \255" }
utf8_spec.rb:4:                        ^~~~~~~~~~~~
Unable to resolve `let` on line 4.

[dgollahon]

@bquorning: Yeah, parser doesn't support non-utf8 strings out of the box so right now I just skip any test files with invalid sequences. I'm pretty sure rubocop has a patch for this or parser has an option to turn it on or something, but I haven't had a chance to enable it yet. Hopefully I can get this stuff working sometime soon. If not, I do want to at least consolidate all the warnings so it just shows one message per file.

[bquorning]

After reading whitequark/parser#323 I don’t think there’s any hope of fixing this issue…

[dgollahon]

Actually, I don't think so. The workaround ought to work just fine for RSpectre. I don't think there are any meaningful cases where I want to do operations on strings that have been parsed out with parser so just letting "invalid" strings fly on through should be fine for us.