-
-
Notifications
You must be signed in to change notification settings - Fork 38
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
New Rules: rune-prefer-let & prefer-const #818
Comments
The code doesn't change the variables, so I think it's fine to use I prefer to use |
@ota-meshi : Although I absolutely can follow your argumentation please keep in mind that this is not what people will intuitively do. Thus if you stick to your approach all tutorials should be adopted and eslint should at least output a message which tells the user to use "const" when working with "$derived". |
I don't force my preferred code style on other people's projects. |
I don't think this issue is a question of code style. Changing code style will never break code. But switching between "let" and "const" does (or at least make the linter not accept the code any more). What I meant: whatever decision the svelte core team mets: please keep in mind to stay consistent here... |
The variable is reassigned by code, but not by your code.
It is not necessary. Using either let or const the Svelte compiler will not allow you to reassign a $derived variable.
In a way, const is not necessary either. When replacing all const statements with let in a codebase, the code will "work" but these variables no longer throw runtime errors when reassigned. But I prefer to use const for variables that are not reassigned.
That the job of the author of a ESLint config, but as a ESLint plugin maintainer you play a rule in forcing a preferred code style. There are two views:
I am in 1, I want Some are in 2 and prefer to use const for $derived because for them it better communicates that this is a readonly value, not necessarily that the value itself never changes and they want to use const for components that only have readonly props. The issue is that regular ESLint only caters to view 2, creates incorrect code for view 1. |
Thank you for explaining, I hadn't thought of it that way. I think it would be a good idea to add rules for users who prefer |
I'm having the same problem in a component with bindable properties. If i set |
Hey @MathiasWP and @ota-meshi, I'll be continuing @bfanger's work in another PR. I think this rule would be helpful in many cases, despite being opinionated and Svelte not strictly defining how it should be done. |
After having worked on this for a bit, how do you feel about adding another property to the rule that would allow devs to prevent reactive declarations from Svelte to be checked by this rule? For instance, I like having my // eslint.config.js
export default [
{
rules: {
'svelte/prefer-const': {
ignoreValues: ['state']
}
}
}
]; So: <script>
// Good, checked and no error
const {} = $props();
// Nothing, rule does not take state into account
const a = $state();
let b = $state();
// Bad, checked and error
let c = $derived(b);
</script> Just throwing this as an idea, which would allow more granular configuration. |
Motivation
const
in javascript:At the moment the Svelte 5 compiler breaks this rule and we can no longer make assumptions of what should be possible.
Because the compiler replaces all usages of reactive values by getters & setters, this doesn't result in a
TypeError: Assignment to constant variable.
error like it would in regular javascript.Description
The svelte/rune-prefer-let rule is aware of which variables are reactive by detecting switch are assigned via a rune.
eslint
prefer-const
rule doesn't see any reassignments so it assumes a let statement can be safely converted toconst
which conflicts with the svelte/rune-prefer-letThe svelte/prefer-const rule is identical, but takes into account which variable are reactive an can be reassigned.
Examples
Additional comments
Implementation of these rules are available #806 & #816
The text was updated successfully, but these errors were encountered: