I don't have principal objection, if you find such hooks useful.

That should not decrease speed, but could you run benchmark for sure? If speed become lower, try to replace if (state.listener) with if (state.listener === null)

/cc @dervus what do you think?

PS. If you executed tests manually, make sure you run linter too (see travis report)

Thanks @puzrin . Delinted. Benchmarks looked very much the same. Seemed slightly better with !== null -- probably negligible and not statistically significant but it made sense so I've done that instead.

LGTM. @dervus any comments about option name and so on?

This change seem to imply public exposure of State object, which I don't like because it would make as to maintain backward comparability for that. I don't object if it'll be non-documented and non-supported feature.

Ok. Let's merge as hidden treasure :)

Released with this PR. Enjoy!