Stop converting to 19 hash syntax when invalid

[crimsonknave]

Version 0.30.0 is now telling me to convert to 19 hash syntax in cases when it should not. There are a few cases where you can have a valid symbol that does not convert to valid 19 hash syntax. According to this comment the regex is pretty straight forward.

Most of the time this would not come up, but I've seen it while setting the [] method in a test double.

failure.rb:2:3: C: Use the new Ruby 1.9 hash syntax.
  :[] => 'asdf'
  ^^^^^^

As requested in the contributing guidelines:

rubocop -V
0.30.0 (using Parser 2.2.0.3, running on ruby 2.0.0 x86_64-linux)

[crimsonknave]

Sorry, I didn't realize there was a test for the changelog (which I added to after I ran the tests). I'm unclear what the test is actually testing and how I can fix it.

[agargiulo]

Figured out what you forgot.

[@username]: https://github.com/username

at the end of the CHANGELOG.md file.

[bbatsov]

This is the changelog for the old version (and this is definitely not a new feature).

[crimsonknave]

I'm not sure why the 1.9.3 tests started printing things after I merged (which appears to be the cause of the failure). I'm not able to reproduce it locally.

@bbatsov how does the less than, greater than filtering look to you? The names are a bit long at the moment, but I felt ruby_lte was a little too dense.

[tfausak]

👍 I would like this patch. I have encountered offenses like this:

example.rb:1:3: C: Use the new Ruby 1.9 hash syntax.
{ :@key => :value }
  ^^^^^^^^

[bbatsov]

The names look ok. The commits should be squashed to just two once you fix the failing build.

[bbatsov]

@crimsonknave Rebase on top of the current master. The second commit message's title should be in present simple tense.

[bbatsov]

👍