That script is attached to a CanvasLayer node with a ProgressBar called HealthBar. And yet, when running the scene, it will throw an error:
This is because there’s actually a spelling error; instead of HealthBar, the node is called HeatlhBar. It’s easy to miss, and we all occasionally make spelling errors.
Since you can easily drag nodes into scripts and create @onready references by holding down CTRL before releasing the mouse button, another version of the script might look like this:
With the script in place, the
health_bar
node can be assigned via the Inspector:
Initially, this may seem like a bit more work, to first have to declare it and then assign it, but it is the most reliable, flexible solution to referencing nodes. You can even rename the HealthBar node, and the
health_bar
node reference will update accordingly.
This approach is especially helpful when building complex user interfaces where you might rearrange nodes a lot. One concern is having a long list of export properties, some of which are actually used to reference nodes internal to the scene. That’s why I typically group them with
@export_group
:
That way, it’s clear in the inspector which properties are meant to be modified from outside the scene:
In addition, this separation between script and scene is a form of dependency injection that provides greater flexibility, so a script can be attached to a scene that has a completely different layout.
I’ve been using this technique almost exclusively, and rarely use
@onready
to assign a node reference via
$
or
get_node()
.
I had a setup with nested CanvasLayer nodes. Toggling the visibility of the root CanvasLayer doesn’t hide any nested CanvasLayer nodes. My solution was to listen to the visibility_changed signal, find any CanvasLayer child nodes, and apply the same visibility to them.
The latest version includes a few new enhancements, and an experiment: The sequencer demo uses inventory instances to hold music notes, which can be played back. This was inspired by music trackers that were popular in the 90s, such as Scream Tracker and Impulse Tracker. The sequencer isn’t meant to be a production-ready digital audio …
The latest version of the inventory system is available and includes refinements for the interaction demo and the main tour. New features: We’ve also got a handful of bug fixes:
Ditch @onready, use @export instead
Are you using @onready to reference nodes? There’s a better way!
Here’s a simple example of how many tutorials use @onready to reference nodes:
That script is attached to a CanvasLayer node with a ProgressBar called HealthBar. And yet, when running the scene, it will throw an error:
This is because there’s actually a spelling error; instead of HealthBar, the node is called HeatlhBar. It’s easy to miss, and we all occasionally make spelling errors.
Since you can easily drag nodes into scripts and create @onready references by holding down CTRL before releasing the mouse button, another version of the script might look like this:
Now it works, because it’s consistently misspelled.
If you now notice the misspelling in the Scene Tree and fix the node name, but don’t adjust the script, it’ll break again:
A better way is to use @export instead:
With the script in place, the
health_barnode can be assigned via the Inspector:Initially, this may seem like a bit more work, to first have to declare it and then assign it, but it is the most reliable, flexible solution to referencing nodes. You can even rename the HealthBar node, and the
health_barnode reference will update accordingly.This approach is especially helpful when building complex user interfaces where you might rearrange nodes a lot. One concern is having a long list of export properties, some of which are actually used to reference nodes internal to the scene. That’s why I typically group them with
@export_group:That way, it’s clear in the inspector which properties are meant to be modified from outside the scene:
In addition, this separation between script and scene is a form of dependency injection that provides greater flexibility, so a script can be attached to a scene that has a completely different layout.
I’ve been using this technique almost exclusively, and rarely use
@onreadyto assign a node reference via$orget_node().Related Posts
Toggling Visibility of Nested CanvasLayers
I had a setup with nested CanvasLayer nodes. Toggling the visibility of the root CanvasLayer doesn’t hide any nested CanvasLayer nodes. My solution was to listen to the visibility_changed signal, find any CanvasLayer child nodes, and apply the same visibility to them.
Inventory System v1.9 available
Another quick inventory system update. The Resizable Container Demo now includes preliminary support for controllers. Features: Additional bug fixes:
Inventory System v1.8 available
The latest version includes a few new enhancements, and an experiment: The sequencer demo uses inventory instances to hold music notes, which can be played back. This was inspired by music trackers that were popular in the 90s, such as Scream Tracker and Impulse Tracker. The sequencer isn’t meant to be a production-ready digital audio …
Inventory System v1.5 available
The latest version of the inventory system is available and includes refinements for the interaction demo and the main tour. New features: We’ve also got a handful of bug fixes: