Minor issues with reconnection behaviour

[cloudhead]

Just a couple of things I've noticed, which are easy to fix:

• When there's a connection error, the client won't attempt to reconnect. This is because the "error" listener doesn't call connection_gone. I'm not sure if this is intentional, but it might be useful if it tried reconnecting on error (such as ECONNREFUSED) after a disconnect.
• client.attempts is never reset. It should probably be set back to 1 in client.on_connect. I'm assuming it's only needed for the exponential backoff, and therefore should be reset after a successful connection.

If you want me to submit a patch, I can do so, but it's really just two lines of code if you want to fix this.

Cheers

[mranney]

I believe this is fixed in v0.6.0.

Thanks for reporting and investigating this. This is a regression, mostly because my test suite doesn't have a good way to simulate all of the different Redis server and connection states.

client.attempts is not used anywhere. It's just a fun counter to see how often you've tried to connect. While looking at this though, I noticed that the retry timer wasn't getting properly reset all the time.