Open-source project
quantum-elixir/quantum-core avatar
quantum-elixir/quantum-core

Quantum's README tracks main, but the newest published release is from February 2024

:watch: Cron-like job scheduler for Elixir

2,418 stars157 forksElixirApache-2.0

At a glance

What is it?
A cron-like job scheduler for Elixir, configured entirely in config/config.exs and started as one supervised child. The gap between the branch it documents and the version on Hex is the interesting part, along with a debug switch that can only be reached by raising the log level for your whole application.
Who is it for?
Judge this on what a scheduler owes you, which is predictable firing and visible failures, and the project is honest about both. The dependency constraint, the three step setup and the job shapes are all spelled out.
Can I use it commercially?
Yes. Apache-2.0 is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
Is it still maintained?
Yes. The repository last received commits 6 days ago.
What is it written in?
Mainly Elixir, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.

From daily snapshots

Star history

  • quantum-elixir/quantum-core2,418 ★ · +2
Star history
Repository2026-09-302026-11-15
quantum-elixir/quantum-core2.4k2.4k

2 Stars in the last 46 days, measured from the daily GitHub snapshot. Compare with other projects

Editorial analysis

The README follows main, and the newest published release is from February 2024

The document opens by warning about itself: it follows main, which may not be the currently published version, and it points readers to hexdocs for the latest published release. The gap is measurable from the release list. The three newest tags are v3.5.1, v3.5.2 and v3.5.3, published on 2024-02-27, 2024-02-28 and 2024-02-29, three consecutive days with no release since. The last commit to the branch is dated 2026-09-22. So roughly two and a half years of work sits on main with no tag attached to it. The setup instruction compounds this: the dependency line asks for ~> 3.0, which admits any 3.x, so a project following these steps can resolve to a published 3.5.3 or to something newer on the branch, and the instructions written here describe main rather than either.

One config namespace holds both the job list and the debug switch

Adding Quantum takes three edits. A module calls use Quantum with an otp_app option pointing at your application, that module is added to the children list of your supervision tree, and the jobs themselves go in config/config.exs. The key point is which name the config is filed under. The scheduler is defined as Acme.Scheduler, but the configuration block is keyed by the otp_app value, so the same :acme namespace that carries jobs: [...] later carries debug_logging: false. Nothing in the walkthrough states what happens when jobs is absent, and it never mentions timezone handling, so a schedule that crosses midnight is left to the reader. The route to more detail is the configuration page on hexdocs, not this document. The examples use a made up :acme application throughout, including in the troubleshooting section, so the key is easy to carry over by mistake.

Quantum's own logs cannot be seen without raising the level for the whole application

The troubleshooting section offers one lever, and it is global:

elixir
config :logger, level: :debug

That switch changes log output for your entire application, not for the scheduler. The document then admits the cost and offers a partial refund, turning Quantum's own messages back off while leaving everything else at debug:

elixir
config :acme, Acme.Scheduler,
  debug_logging: false

So there is no per job or per call verbosity, only the application wide level plus a single filter for the scheduler. The practical consequence lands on the next instruction, which tells you to search for an existing open issue and, failing that, to open a new one with debug logs attached. Producing those logs means lowering the level for your whole system first. Anyone running this in production has to either accept that or run with the level raised, which is the trade the two config lines encode.

Job definitions accept MFA tuples, anonymous functions, and an @daily alias

The four example jobs use three different shapes, which is the clearest statement of what the scheduler will run:

elixir
config :acme, Acme.Scheduler,
  jobs: [
    # Every minute
    {"* * * * *",      {Heartbeat, :send, []}},
    # Every 15 minutes
    {"*/15 * * * *",   fn -> System.cmd("rm", ["/tmp/tmp_"]) end},
    # Runs on 18, 20, 22, 0, 2, 4, 6:
    {"0 18-6/2 * * *", fn -> :mnesia.backup('/var/backup/mnesia') end},
    # Runs every midnight:
    {"@daily",         {Backup, :backup, []}}
  ]

A module, function and argument list is given as a tuple. A one line shell call is given as an anonymous function. An Erlang function call sits inside a third anonymous function. The schedule field mixes the familiar five field cron syntax with a named alias, since @daily appears beside * * * * * in the same list. Two of the four examples compute a value that nothing in the document consumes: System.cmd returns a tuple and :mnesia.backup returns its own result, and neither is bound to anything.

The overnight range wraps past midnight, and the syntax is never explained

One expression in the list is the only place the extension shows. The schedule 0 18-6/2 * * * is annotated as running on 18, 20, 22, 0, 2, 4 and 6, which is a range starting at 18 in the evening, stepping by 2, passing the end of the day, and continuing to 6 in the morning. The standard five field cron grammar most people arrive with does not wrap a range past its end, so this line is doing something the reader has to notice rather than read. The comment states which hours fire and nothing else: it does not say that the range wraps, does not mention the step, and does not say what happens when the start is larger than the end in a five field field of ordinary cron. The document offers no example of that case, so the behavior for anything but this one line is left to the configuration page on hexdocs.

The scheduler is one supervised child under :one_for_one

The wiring is a standard OTP application, and the scheduler is added as an ordinary child:

elixir
defmodule Acme.Application do
  use Application

  def start(_type, _args) do
    children = [
      # This is the new line
      Acme.Scheduler
    ]

    opts = [strategy: :one_for_one, name: Acme.Supervisor]
    Supervisor.start_link(children, opts)
  end
end

The strategy is :one_for_one and the module is the only child in the example, so a scheduler that exits is replaced without touching anything else in your tree. Nothing in the document states what a crashed job does to that child, whether a failing job takes the scheduler down with it or is isolated, and nothing gives a restart limit or backoff. The example also carries no job list at all, which makes the point that the module can be started before any job is configured. Root entries include a .tool-versions file and a dialyzer.ignore-warnings file, so a toolchain is pinned and a static analysis warning list exists, yet no Elixir or OTP version is stated anywhere in the setup.

Copyright 2015 under an organization repository, with mandated issue titles

The license block reads Copyright (c) 2015 Constantin Rack and grants use under the Apache License, Version 2.0, while the contribution guide asks contributors to fork the quantum-elixir repository on GitHub. A single name and a single year sit under an organization namespace. The contribution process is a seven step list built on the Collective Code Construction Contract, and two steps are naming requirements rather than engineering ones: issues must be titled Problem: followed by a description, and pull requests must be titled Solution: followed by a description. Step one also sends the reader to a linked video before any discussion begins, and step seven promises that a merged change adds you to the contributors graph. Between the org hosting and the mandated prefixes, the document expects a good deal of process for what is a small library.

Editorial conclusion

Judge this on what a scheduler owes you, which is predictable firing and visible failures, and the project is honest about both. The dependency constraint, the three step setup and the job shapes are all spelled out. What it does not spell out is the version you are reading: the page follows a branch whose newest release is v3.5.3 from 2024-02-29 while commits have landed as recently as 2026-09-22. Pin an exact version before you rely on a schedule, and read the configuration page on hexdocs rather than trusting the branch copy, because the two can describe different behavior.

Frequently asked questions

How do I add Quantum to an Elixir project?

Three edits. Add {:quantum, "~> 3.0"} to the dependency list in mix.exs, define a module that calls use Quantum with an otp_app option, and put that module into the children list of your application supervision tree. Jobs are then configured separately.

Where does Quantum read its cron jobs from?

From config/config.exs, in a block keyed by your application's otp_app name and the scheduler module, under a jobs: option holding a list of schedule and job pairs. Schedules may be five field cron expressions or named aliases such as @daily.

How do I silence Quantum's own debug messages?

Set debug_logging: false for your scheduler module in the same configuration block where jobs is defined. Raising the log level is still required to see them, because the instructions tell you to set the global logger to :debug first.

What license is Quantum released under?

The Apache License, Version 2.0. The license block carries a copyright line naming Constantin Rack and the year 2015, and a LICENSE file sits at the repository root.

Which Elixir and OTP versions does Quantum's setup target?

The setup steps name no Elixir or Erlang version and give no platform requirement. The repository root does carry a .tool-versions file, so a toolchain is pinned somewhere the walkthrough never shows.

Official sources

  1. License: Apache-2.0
  2. Project website
  3. quantum-elixir/quantum-core on GitHub
  4. README
  5. Releases
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/quantum-elixir-quantum-core.svg)](https://hysenlabs.com/projects/quantum-elixir-quantum-core)

Star history chart for your README

markdown
[![Star history](https://hysenlabs.com/chart/quantum-elixir-quantum-core.svg)](https://hysenlabs.com/projects/quantum-elixir-quantum-core)

Was this useful?