Fix for camelCase and unused argument name conflict

[mmcguinn]

Previously, the the behavior was something like the following:

foo      -> snake_case or camelCase
foo_bar  -> snake_case
fooBar   -> camelCase
_foo     -> unused snake_case
_foo_bar -> unused snake_case
_fooBar  -> invalid

New behavior should be:

foo      -> snake_case or camelCase
foo_bar  -> snake_case
fooBar   -> camelCase
_foo     -> unused snake_case or camelCase
_foo_bar -> unused snake_case
_fooBar  -> unused camelCase

Please let me know if there an appropriate place to add a test(s) for this and I'd be happy to add them. I looked at the existing tests and it doesn't look like they test any integration like this (and this change passes existing unit tests).

[mmcguinn]

Forgot to explicitly link to the issue, #2758

[alexdowad]

Please let me know if there an appropriate place to add a test(s) for this and I'd be happy to add them.

There should always be tests.

Which was the cop which was (erroneously) failing before? You could add a test case for that cop, which confirms that the code which previously failed doesn't fail any more.

[mmcguinn]

I was originally thinking that a combination the Style/VariableName cop and the relevant Lint/UnusedMethodArgument, Lint/UselessAssignment or Lint/UnusedBlockArgument cop was responsible (which is why I was unsure where to put tests). Now that I come back to it though it doesn't really have anything to do with the lint cops directly. I'll add a couple tests to the VariableName cop.

[mmcguinn]

Okay, I've added a pair of tests to the variable name cop and squashed all the commits down to one.

[bbatsov]

Looks good. Just update the commit message to match our commit messages style - [Fix #n] Add optional .... See http://chris.beams.io/posts/git-commit/ for details.

[bbatsov]

Would you mind rebasing this one more time? I didn't see your update and merged another PR before this one.

[mmcguinn]

No problem - should be good to go once Travis confirms it again.

[bbatsov]

👍