fix: exclude readonly bindings from reactive state validation

[dummdidumm]

bind:this etc only go upwards, and depending on the use case it's fine to connect it to a non-reactive variable

Before submitting the PR, please make sure you do the following

• It's really useful if your PR references an issue where it is discussed ahead of time. In many cases, features are absent for a reason. For large changes, please create an RFC: https://github.com/sveltejs/rfcs
• Prefix your PR title with feat:, fix:, chore:, or docs:.
• This message body should clearly illustrate what problems it solves.
• Ideally, include a test that fails without this PR but passes with it.

Tests and linting

• Run the tests with pnpm test and lint the project with pnpm lint

[changeset-bot]

🦋 Changeset detected

Latest commit: be331d7

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

This PR includes changesets to release 1 package

Name	Type
svelte	Patch

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

[Rich-Harris]

Is bidirectionality the right criterion? Feels a bit weird to exclude <svelte:window bind:innerWidth={pojo.value}> — if I'm not reacting to changes in innerWidth then what's the point of the binding?

If we want a basis on which to exclude bind:this, then something like 'is only updated once' would surely make more sense (though it would need to be combined with a 'is not in a block' check).

Side-note: I thought we were already warning on non-reactive Identifier bindings at compile time, but it seems that we're doing no such thing. Did I hallucinate that or did we stop doing it?

[dummdidumm]

Is bidirectionality the right criterion? Feels a bit weird to exclude <svelte:window bind:innerWidth={pojo.value}> — if I'm not reacting to changes in innerWidth then what's the point of the binding?

<script>
  let online;
  let innerWidth;

  function printStuff() {
    if (!online) alert('cant do this right now')

    let size = innerWidth / ...;
    createPicture(size);
    print();
  }
</script>

<svelte:window bind:online bind:innerWidth />

Similar examples can be created for the other readonly bindings.

Side note: the above currently creates a compiler warning, is that correct? I feel like it's a bit overzealous.

If we want a basis on which to exclude bind:this, then something like 'is only updated once' would surely make more sense (though it would need to be combined with a 'is not in a block' check).

This PR started out as "don't warn on bind:this" and then I thought of the more general one-way-bindings and felt it should apply there, too.

Side-note: I thought we were already warning on non-reactive Identifier bindings at compile time, but it seems that we're doing no such thing. Did I hallucinate that or did we stop doing it?

We are warning on it?

[trueadm]

We don't warn on this

[dummdidumm]

Yeah but that's correct, because it doesn't need to be reactive, it would be (was! we changed this) very annoying to have a warning in this case.

[trueadm]

@dummdidumm I mean, I'd argue that it should be though. It keeps it consistent with the rest of our reactivity game.

[Rich-Harris]

I would write the above example like this:

<script>
  function printStuff() {
    if (!navigator.onLine) alert('cant do this right now')

    let size = innerWidth / 1;
    createPicture(size);
    print();
  }
</script>

The only reason to use a binding is because you want something to be reactive. bind:this is a bit of a special case because it updates exactly once, and it does so immediately before the first effects run (e.g. onMount). I don't think there's any other binding with those characteristics.

[dummdidumm]

Adjusted it so it's only silencing the "bind:this not in a block" case