All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog and this project adheres to Semantic Versioning.
- don't mimic reply for commands
- syntax issue when matching nicks
- Exception logging for generate_models in separate thread
- build markov file for channel
- bad indent led to cobe responses to not output if not replacing matched aliases
- when replying to a ping, address the nick who pinged, not another nick
- fix regression where cobe brain wasn't being built on command
- chain all alias markov models for associated nick
- use seperate thread for markov models
- fix regression in 'no models found' handling
- fix regression in regex handling message exclusions being ignored
- DRY markov file generation
- fixed filename for externally loaded models
- store markov chains as files
- subcommand for tweeted last response
- using Plugin subclass
- subcommand for loading external corpora
- setting for cobe reply time
- improved corpus filter
- added a function for filtering messages into the markov chains
- use helga-alias to generate more complete models
- support for ignore list setting
- post processing of cobe generated replies
- use cobe for channel-wide mimic, much faster
- added a method for other plugins to use helga-mimic to generate replies
- added cobe module to make some conversation
- added a temporary command ("!mimic build") to rebuild the brain
- cleaned up some logging
- fixed some unit tests
- no argument suppplied means to mimic current channel
- can specify multiple nicks
- I suck at deploying
- setting to allow to change number of tries
- increased number of tries from 3 to 50
- allow anyone to call mimic
- allow operators to generate mimics based on oral history