Chaostoolkit

Changed

- Publish events for each step of the flow

Changed

- New `chaos init` wizard instructions

Changed

- Returning journal and experiment from run and validate commands for
downstream applications
- Better guidance on init

Changed

- Create an empty experiment when no discovery was run beforehand [27][27]
- Returns the generated experiment to external callers
- Name of the journal output from the run command is now `journal.json` rather
than `chaos-report.json` [31][31]
- Renamed the debug log from `experiment.log` to `chaostoolkit.log` because
it is used for any commands, even when the experiment is not required
- The debug log is now appending
- The command being run is logged into the debug log
- You can bypass argument in the init command via empty string [29][29]
- Allow to create steady-state hypothesis from init command [28][28]
- Allow to set rollbacks from init command [30][30]
- Pass command executed to checker for compatability [36][36]
- Better logging of failed discovery [chaostoolkit-lib29][29lib]
- Depending now on chaostoolkit-lib 0.14.0

[27]: https://github.com/chaostoolkit/chaostoolkit/issues/27
[28]: https://github.com/chaostoolkit/chaostoolkit/issues/28
[29]: https://github.com/chaostoolkit/chaostoolkit/issues/29
[30]: https://github.com/chaostoolkit/chaostoolkit/issues/30
[31]: https://github.com/chaostoolkit/chaostoolkit/issues/31
[36]: https://github.com/chaostoolkit/chaostoolkit/issues/36
[29lib]: https://github.com/chaostoolkit/chaostoolkit-lib/issues/29

Added

- Steady state hypothesis is not mandatory when exploring weaknesses [18][18]

[18]: https://github.com/chaostoolkit/chaostoolkit/issues/18

Added

- New init feature [23][23]

[23]: https://github.com/chaostoolkit/chaostoolkit/issues/23