In other frameworks it’s common for components to perform side effects whenever the need them. An event handler might make an HTTP request, or a component might reach into the DOM to focus an input.

In Lustre we try to keep side effects separate from our main program loop. This comes with a whole bunch of benefits like making it easier to test and reason about our code, making it possible to implement time-travel debugging, or even to run our app on the server using Lustre’s server components. This is great but we still need to perform side effects at some point, so how do we do that?

The answer is through the Effect type that treats side effects as data. This approach is known as having managed effects: you pass data that describes a side effect to Lustre’s runtime and it takes care of performing that effect and potentially sending messages back to your program for you. By going through this abstraction we discourage side effects from being performed in the middle of our program.


For folks coming from other languages (or other Gleam code!) where side effects are often performed in-place, this can feel a bit strange. A couple of the examples in the repo tackle effects:

This list of examples is likely to grow over time, so be sure to check back every now and then to see what’s new!

Getting help

If you’re having trouble with Lustre or not sure what the right way to do something is, the best place to get help is the Gleam Discord server. You could also open an issue on the Lustre GitHub repository.

While our docs are still a work in progress, the official Elm guide is also a great resource for learning about the Model-View-Update architecture and the kinds of patterns that Lustre is built around.


The Effect type treats side effects as data and is a way of saying “Hey Lustre, do this thing for me.” Each effect specifies two things:

  1. The side effects for the runtime to perform.

  2. The type of messages that (might) be sent back to the program in response.

pub opaque type Effect(msg)


pub fn batch(effects: List(Effect(a))) -> Effect(a)

Batch multiple effects to be performed at the same time.

Note: The runtime makes no guarantees about the order on which effects are performed! If you need to chain or sequence effects together, you have two broad options:

  1. Create variants of your msg type to represent each step in the sequence and fire off the next effect in response to the previous one.

  2. If you’re defining effects yourself, consider whether or not you can handle the sequencing inside the effect itself.

pub fn from(effect: fn(fn(a) -> Nil) -> Nil) -> Effect(a)

Construct your own reusable effect from a custom callback. This callback is called with a dispatch function you can use to send messages back to your application’s update function.

pub fn map(effect: Effect(a), f: fn(a) -> b) -> Effect(b)

Transform the result of an effect. This is useful for mapping over effects produced by other libraries or modules.

Note: Remember that effects are not required to dispatch any messages. Your mapping function may never be called!

pub fn none() -> Effect(a)

Most Lustre applications need to return a tuple of #(model, Effect(msg)) from their init and update functions. If you don’t want to perform any side effects, you can use none to tell the runtime there’s no work to do.

Search Document