Not sure if this is optimally clean code, it could certainly use some DRYing up, but I'm not enough of a wizard with the codebase to see the refactoring path just yet. Feel free to clean it up.

The code looks fine to me, but it seems you don't actually do anything with the Hash, right? Perhaps we can add in the little bit of extra logic that gets the encoding from the hash?

The options hash is the same options hash provided to the other method variants (String#encode(encoding, opts)). The options hash doesn't have an encoding option, only options with which to perform conversion to the Encoding.default_internal encoding. See the docs: http://rubydoc.info/stdlib/core/String:encode

Note that one thing I noticed we might be missing is that: "The last form by default does not raise exceptions but uses replacement strings. " -- this seems to imply that the default value for the :invalid option is :replace. Is that how you interpret it?

On most text-encoding systems I've dealt with, "replace" is usually the default for invalidly-encoded characters.

I'll go ahead and pull in your change today and we can improve support through other revs.

On second thought, since we have a release going out today, let's defer this to 1.6.2 and we'll try to fill out the functionality a bit more...

We should revisit this issue as the 1.6.2 release draws near.

Bumpy bump.