Skip to content
  • Nathan Harris's avatar
    Resolve PubSub post-commit feedback · e858c0a6
    Nathan Harris authored
    Motivation:
    
    To ship PubSub faster, it was merged to the master branch without a peer review. This commit is to address the critical points of feedback given in a post-commit review.
    
    Modifications:
    
    - Add: New RedisClientError case where a "race condition" of removing a pubsub handler and subscription can happen
    - Add: New state to RedisPubSubHandler for when it has been removed from a ChannelPipeline
    - Change: RedisPubSubHandler to require an `eventLoop` in its initializer
    - Change: The subscribe and unsubscribe methods on RedisPubSubHandler to handle the EventLoop hopping to be thread-safe
    
    Result:
    
    PubSub should have a more robust and thread-safe implementation.
    e858c0a6