Throttle messages for a slow consumer¶

This example shows how to throttle messages that are published are a rate faster than what a slow consumer (subscribed) can process. In this example, only the newest messages are preserved, messages that cannot be consumed on time are dropped.

import time
import logging

from roslibpy import Header
from roslibpy import Ros
from roslibpy import Time
from roslibpy import Topic
from roslibpy.core import LOGGER

# Configure logging
fmt = '%(asctime)s %(levelname)8s: %(message)s'
logging.basicConfig(format=fmt, level=logging.INFO)
log = logging.getLogger(__name__)

client = Ros(host='', port=9090)

def receive_message(msg):
    header = Header(msg['seq'], msg['stamp'], msg['frame_id'])
    age = time.time() - header['stamp'].to_sec()
    fmt = 'Age of message (sequence #%d): %6.3f seconds', msg['seq'], age)
    # Simulate a very slow consumer

publisher = Topic(client, '/slow_consumer', 'std_msgs/Header')

# Queue length needs to be used in combination with throttle rate (in ms)
# This value must be tuned to the expected duration of the slow consumer
# and ideally bigger than the max of it,
# otherwise message will be older than expected (up to a limit)
subscriber = Topic(client, '/slow_consumer', 'std_msgs/Header',
                            queue_length=1, throttle_rate=600)

seq = 0
def publish_message():
    global seq
    seq += 1
    header = Header(frame_id='', seq=seq,
    client.call_later(.001, publish_message)


In the console, you should see gaps in the sequence of messages, because the publisher is producing messages every 0.001 seconds, but we configure a queue of length 1, with a throttling of 600ms to give time to our slow consumer. Without this throttling, the consumer would process increasingly old messages.