feat: add excludedRunes option to the prefer-const rule

[baseballyama]

close part of #818

As explained here, $state also needs to be excluded from the prefer-const rule.
#985 (comment)

[changeset-bot]

🦋 Changeset detected

Latest commit: e4f255e

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
eslint-plugin-svelte	Minor

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[github-actions]

Try the Instant Preview in Online Playground

ESLint Online Playground

Install the Instant Preview to Your Local

npm i https://pkg.pr.new/eslint-plugin-svelte@1064

Published Instant Preview Packages:

• eslint-plugin-svelte: https://pkg.pr.new/eslint-plugin-svelte@1064

View Commit

[ota-meshi]

I still don't understand why you want to ignore $state from svelte/prefer-const rule.

I know that some users want to use let to change value with const value = $state({ value: new Date().toString() });, but I think that user would do the same with const value = { value: new Date().toString() }; which doesn't use $state(). In other words, I think that user would always turn off the prefer-const rule.
So I don't think there's a need to ignore $state from the svelte/prefer-const rule check.

[baseballyama]

We might want to enable prefer-const for plain JavaScript that has nothing to do with Svelte.

In other words, semantically, since $state is something that can be reassigned, we may want to consistently use let for it. However, it’s understandable that some users would prefer to use prefer-const for regular JavaScript outside of that.

(Personally, I always use const consistently, though.😅)

[baseballyama]

In this case, there is no actual reassignment, so prefer-const would allow it to be declared as const. However, semantically, some people may prefer to declare it with let.

export const Time3 = () => {
	// If `$state` has object value,
	// Time3 user cn reassign `value.value` even without reassign logic in this composable.
	// Therefore it can be declare as both `const` / `let`
	const value = $state( { value: new Date().toString() });

	return {
		get value() {
			return value;
		}
	}
}

[ota-meshi]

I understand that when a user uses $state(), it is clear that the user intends a variable that will be mutated. However, objects that do not use $state() may also be mutable, so for consistency, I think that users who prefer let should also use let for them.
In other words, the prefer-const issue with mutable objects is not a problem specific to $state() (Svelte syntax), so I don't think it should be specially ignored.

[baseballyama]

Therefore, I am considering creating a separate rule, prefer-let. This rule will enforce the consistent use of let.

[baseballyama]

@ota-meshi So, what you’re saying is that extending prefer-const in the first place is unnecessary?

[ota-meshi]

So, what you’re saying is that extending prefer-const in the first place is unnecessary?

No, I think svelte/prefer-const is necessary.
I don't think there's any need to ignore $state() from the svelte/prefer-const rule check.

Since we don't ignore svelte/prefer-const rule checks for mutable objects regardless of $state(), I think the rule would be inconsistent if we were to ignore only $state() from svelte/prefer-const rule checks.

If a primitive value of $state() is reported with prefer-const, it's just a case of the user using it incorrectly, and I think we can follow it with a different rule (e.g. prefer-let).

[baseballyama]

If a primitive value of $state() is reported with prefer-const, it's just a case of the user using it incorrectly, and I think we can follow it with a different rule (e.g. prefer-let).

I still don’t fully understand this part. Since $state can hold non-primitive values such as object, are there cases where there is no direct reassignment? In such cases, is it correct for const to be enforced? Even if a prefer-let rule were implemented, wouldn’t it conflict with prefer-const, which enforces const?

https://svelte.dev/playground/hello-world?version=5.19.9#H4sIAAAAAAAAE32MzQqDMBAGX2VZCiqIlh6DFXrrO9Qeom4lkMaQrLZFfPcSpT-H0tsy38xOaOSVUOCRtO7h1jvdQkytYmoTTPGiNHkUpwn5YYMXAKav6mBt5kfSHFgtPf3iTW-YDHsUWPjGKctlZSrWxDB4crCHjWfJFMMEoRcQSa0ailKQHQnYbWGGpDJF_qlNYcvVncKPLNxzkdtyWZZsHWT35srYgaFWphWj1APtv1LI_xjhxyJgikx3RsFuoPk8PwFxCBbbPgEAAA==

[ota-meshi]

If in that example, user should be declared with let, then I think user should also be declared with let in the following source code.
What do you think?

https://eslint-online-playground.netlify.app/#eNq1VNuO0zAQ/RXLQtq2apMCb9nuConLG4J3AlJIJpVL4li2U3VV9d8Z3xK3CQsS2j7U8cw5x3Ozz1TJMv389L5rRceB60QdodFAM7pTpWRCP+ackAY04UUL5IG8UrrQsLgrGlbC3TJ4i33kfLNFe66NvVcg0XE2OEL2XmexDBZCJOhecmt2psvarSqAj0XTQ8TwkVjzNcXoYyQz8midqhvorbhLJNY2C/7t0qEguBGPJoqMnE2Cifm+7FKBPvSgRHDgZ7AzLnpNfjJeZVb9IaKS9BmE0bAAuqagGoZNKjtes31yUNgnhp2TmhwUqWXXkpy+c6D0oHJ6n3Pvd3392vR7xgPSATfCGjcOMcMppGmi5zjbJlCtL+Zo5VwBr58EuMJ5jgXDyYIrqIu+GUk+sYUp+UH5nUoklF3bAq+gst1LkuSGMYuJUw64bzmtm0KnETyn3y3Dj0DNGlAZQeBqla68SE7XmMu4cwwc/YLve2zQF6FZx5E2zJGrTHZVQ08KzilppEUmguXJTF8npUxD+Qfhcb7ND05aFp8wn48nDVy5wzCxSRojLayyt1UYwgiNT4WEGuQGy6mwlSYukLLDGOLb8td6Yv+uS2oNL17VYWqcYZr9P0ZvrtZV9MbwX9EPJy/v8ZqLovyFCijbcbziVsBdJVfytquwP77vOa3g+AGEGWZeMsBQhjOvXwNDxeEH0zl/bngEnnXevhAG+uNtsk22m20Ejt0Tpfl3wyq9niiNk/4HtZlXJcb5J/tCL78B2OY91w==

[ota-meshi]

The area I have an issue with is the consistency of the rule.

[baseballyama]

I fully understand your perspective.👍

As stated in the documentation, Svelte runes recommend using let. Based on my understanding, some users interpret JavaScript rules as being overridden by Svelte rules.
https://svelte.dev/docs/svelte/$state

In your example, since user does not use runes, I understood that standard JavaScript rules apply (without being overridden by Svelte rules). Therefore, I believed that the behavior of this PR was consistent.

Additionally, based on your reasoning, shouldn’t $props and $derived also consistently use const whenever possible? (I may not yet fully understand this part.)

[ota-meshi]

I think the reason why $derived should be let is because it is changed from outside.
But $state doesn't really fit that reason.
For primitive values, the user should assign it somewhere, so it won't be reported to prefer-const.
If the reason to use let is that the object is mutable, then I think objects with setters should use let.

Svelte runes recommend using let.

Hmm... I don't think it explicitly says "recommended". I think it's still a matter of user preference 🤔

[baseballyama]

I see. Would adding a config be a good middle ground? Like { enforceState: false }?
I want to accommodate users who wish to use prefer-const in some way but also want to consistently use let for runes.

[ota-meshi]

I want to accommodate users who wish to use prefer-const in some way but also want to consistently use let for runes.

If there are users requesting it, I think it makes sense to add the option.

[baseballyama]

Got it. In that case, since there are no breaking changes, I will try implementing this option after the v3 release.

[baseballyama]

I added excludedRunes option.