-
Notifications
You must be signed in to change notification settings - Fork 182
SocketScience
SocketScience is currently disabled.
SocketScience is SmokeDetector's internal library to allow for direct communication between instances. We use this instead of a more obvious solution like WebSockets because it doesn't require any special considerations or configuration from hosts: whereas WebSockets would require hosts to open a port, SocketScience uses chat - which every instance is already connected to.
SocketScience has two primary use-cases:
- Send some data to other instances
- Do something with data that another instance has sent
These are, respectively, covered by the send
and register
methods.
In short: call SocketScience.send(payload)
.
payload
should be a dict
. The keys of this dict are message types - for example, you might have one message type for metasmoke up/down reports (metasmoke_state
), and another for Smokey pings (ping
). Do not use top-level dict keys for properties of your message; instead, assign another dict to the key for the relevant message type in the top level dictionary.
Bad:
SocketScience.send({'location': GlobalVars.location, 'time': time.time()})
Good:
SocketScience.send({'ping': {'location': GlobalVars.location, 'time': time.time()}})
This construct is important, because it's what the receiving end relies on to work.
When SmokeDetector starts, you should register a callback with SocketScience for the message type you want to respond to. To do this, call SocketScience.register(prop, cb)
- prop
is the relevant message type, and cb
is a method that SocketScience will call when a message of that type is received.
For example, if you want to respond to ping events by telling Charcoal HQ which instance pinged and when (this is a terrible idea and I will hunt you down if you do it, but it serves as an example):
def handle_ping(data):
chatcommunicate.tell_rooms_with('debug', '{} pinged at {}'.format(data['location'], data['time']))
SocketScience.register('ping', handle_ping)
As that example indicates, your callback will be passed a single parameter, containing the contents of the data dict for that message type only.
Say you want to create a new message type beverages
that makes every instance serve a user a !!/coffee
or !!/tea
instead of just one (this is also a terrible idea).
# ws.py (most of file omitted)
def serve_beverage(data):
chatcommunicate.tell_rooms_with('debug', ':{} *serves another {}*'.format(data['message_id'],
data['beverage_type']))
SocketScience.register('beverage', serve_beverage)
# chatcommands.py (relevant sections only)
@command(whole_msg=True)
def coffee(msg):
# rest of command omitted
SocketScience.send({'beverage': {'message_id': msg.id, 'beverage_type': 'coffee'}})
@command(whole_msg=True)
def tea(msg):
# rest of command omitted
SocketScience.send({'beverage': {'message_id': msg.id, 'beverage_type': 'tea'}})
Smokey is made with <3 by the Charcoal Team and the other awesome contributors from Charcoal HQ.