Skip to content

Device Modules

Stephen von Takach edited this page Mar 16, 2017 · 4 revisions

Device drivers implement a protocol on a raw transport stream, TCP or UDP. For examples see our open repository of device modules

Some of the configuration is defined in the database or in backoffice.

image

  • Transport Layer Security (TLS): Should communications be encrypted
  • UDP: Should we be stateless (UDP connections are always considered connected)
    • Multicast IPs are automatically detected and device modules will be subscribed to the multicast group
  • Make Break: Do we want the module to disconnect when there is nothing to communicate
    • Some devices require you work this way

Typical Layout

class Manufacturer::Type::ModuleName
  # Called on module load before the connection is initiated
  def on_load; end

  # Called before the module is cleaned up (disconnected already)
  def on_unload; end

  # Called after dependency reload and settings updates
  def on_update; end

  # Called when the TCP connection to the device is completed
  def connected; end

  # Called when the connection to the device is lost
  def disconnected; end

  # Called when data is sent from the device
  def received(data, deferrable, command)
    # data == data from the device as a String
    # deferrable == allows you to decide if a result was a success asynchronously
    # command == last command sent that hasn’t been resolved (raw command + metadata)
  end
end

Transmitting a request

Requests are added the transmit queue by calling send(raw_cmd, options_hash).

The raw_cmd can be in a number of formats:

  • String => will be transmitted as is
  • Array of bytes => will be automatically converted to a string for sending
  • Hex String => will be converted to a binary string if requested send('0xbeef', hex_string: true)

The options are as follows:

Option Default Value Description
wait true do we want to wait for a response before we continue processing
delay 0 minimum delay time between sends (milliseconds)
delay_on_receive 0 time to delay the next transmit after receiving data (milliseconds)
max_waits 3 number of times we'll accept an ignore response before retrying the request
retries 2 number of times we'll retry a command if it has failed
timeout 5000 amount of time we'll wait for a response to a command before retrying (milliseconds)
priority 50 so we can perform commands in preference to others (see section on priorities)
force_disconnect false causes the transport to disconnect once a response has been received
clear_queue nil/false removes any other queued commands once it starts transmitting
emit nil callback to occur when that request completes. Will not be called if another request with the same name overrides this request.
on_receive nil alternative receive function or block. Called in stead of received function

Processing a response

It's expected that the received function or on_receive callback return a result for the current command. When data is received, this function is called to see if

  • The command was a success
  • This data has nothing to do with the command in question
    • Some devices push data
  • Some kind of failure

The expected values that should be returned by this function:

Value Description
true, :success (or result if not a below value) We've finished processing this command, move on to the next
false, :retry, :failed, :fail The command didn't have the desired result, maybe the device was busy. Please send this command again
nil, :ignore This data was not in response to our action. Continue waiting.
:abort The command failed and it should not be retried. Abort differs from success as it is logged.
:async Waits for the command to resolved by the deferrable passed to the received function

Priorities

Priorities ensure requests are processed in a sane order. For example if you are polling a projector for status and it only accepts one command every 300ms you don't want to wait for the polling, which might be 4 or 5 requests, to complete before executing a more important request like selecting a new input.

Sometimes a query might be made for control flow purposes so it is often useful to differentiate between user initiated requests and polling.

# A best practice query function
def power?(opts = {}, &block)
    opts[:emit] = block if block_given?
    opts[:name] = :power_query
    send('power_query', opts)
end

# Example polling function
def poll
    power? priority: 0 do
        if self[:power] == On
            input? priority: 0
            volume? priority: 0
        end
    end
end

Priorities are also increased on a contextual basis. This is how retries, for instance, make their way back to the front of the queue - which is what you would expect. There is a configuration option called priority_bonus which increases the priority of a command in the following circumstances:

  • Any request made when processing a response to a request
    • If you request volume status whilst processing the switching of an input, the query will receive a bonus and jump towards the front of the queue
  • When a command fails and is retried, it also receives a bonus.

NOTE:: given default values, if you send commands in the received function and then return failed so the command is retried, the commands sent will be executed before the retry. You can counter this effect by providing lower priorities in this context.

Helper functions

Name Description
disconnect disconnects the current connection. It does not wait to send any buffered data
remote_address returns the IP address or hostname defined in the database
remote_port returns the port number defined in the database
defaults(options_hash) allows you to set custom default options for commands
config(options_hash) allows you to set custom processing configs
set_connected_state(true false)

For the various defaults and config options see the command processor.

When you include ::Orchestrator::Constants some common configuration and default options are exposed in a more declarative manner.

tokenize delimiter: "\xAA"  # See the page on Tokenisation
delay between_sends: 200, on_receive: 100
wait_response false
queue_priority default: 50, bonus: 20
clear_queue_on_disconnect!
flush_buffer_on_disconnect!
before_transmit :run_function

def run_function(data, command)
    # You can modify the data at the last min here (might be waiting in the queue for awhile)
    return data
end

# For make break connections, in milliseconds
inactivity_timeout 5000