If the first data passed to Diff::LCS::Hunk is empty then we get an error due to the encoding detection introduced in 1.2.2.
This attempts to use the old data as an encoding source, falls back to the new data as an encoding source and then finally to a string literal.
protect against no data
fix indentation of examples
I'll merge this as soon as I file a bug with Rubinius for the failed build in 1.9 mode—I need to be able to make a test case for them to see the problem in a somewhat better way. I'm treating it similar to the MacRuby issue; everyone else gets it right, why doesn't rbx?
That should be by the afternoon, and I'll do a release this evening.
Hey thanks! Could you ping me (via mention or whatever) when you've rolled a release? We're awaiting it over on RSpec :)
@JonRowe the release is complete.
Thanks @halostatue! Appreciate it :)