Recurring

Build Status Coverage Status Documentation Status

This is a simple library for running a function or callable every N seconds. It’s meant for applications that need to schedule small, self-contined callable(s) on a relatively long, potentially changing period . alive-checks, state snapshots, that sort of thing.

Use this if:

  • You want to call something periodically over the lifetime of your application.
  • You want to be able to change the time between calls.
  • You want or need to avoid the overhead of joining and starting a thread every time. (up to 1/5 of a second according to my sample-size of one machine under no other load)
  • The stuff you’re going to call isn’t going to destroy machines if it’s killed abruptly at the end of the application’s life.

This is probably not appropriate for your project if:

  • You’re already using or likely will be using a fleshed-out concurrency framework.
  • You have many things you’d like to repeatedly schedule and run.
  • Your callables absolutely must execute some cleanup code to avoid disaster on kill.

This is not a library intended for top-level program composition.

Installation:

pip install recurring

Usage:

import recurring

def stuff():
    # do stuff ...

seconds_between_stuff = 30

job = recurring.job(stuff, seconds_between_stuff)
job.start()

# ...

seconds_between_stuff = 300000000 # this will be *from when rate is set*, not *from the next scheduled call*
job.rate = seconds_between_stuff

# ...

# stop making calls until start() is called again
job.stop()

# some time later ....
job.start()

# stop making calls permanently
job.terminate()
job.start() # raises RuntimeError
job.rate = 3000 # raises RuntimeError

Changelog

2.0.0 - 2018-05-30

  • replaced sched backend with threading.Timer-like implementation, saving us from needing to respawn when a job’s rate is changed.
  • jobs can now be permanently stopped by calling job.terminate()

Backwards-Incompatible Changes

  • job.stop() no longer takes an optional timeout argument

1.0.1 - 2018-05-24

  • Corrected an assumption about the number of events that could be queued at once.

1.0.0 - 2018-05-22

  • Initial release

Indices and tables