linter: `eslint/no-inner-declarations` does not report for `var` in for loop initializer

[overlookmotel]

What version of Oxlint are you using?

Latest main

What command did you run?

No response

What does your .oxlintrc.json config file look like?

No response

What happened?

The following cases are not considered an error under eslint/no-inner-declarations rule:

for (var x in {}) {}
for (var x of []) {}
for (var x; x < 10; x++) {}

ESLint considers all these to be errors (ESLint playground). Is there a reason why we don't?

[camc314]

@overlookmotel is this fixed by #11632 ?

I assigned to you. Do you mind closing if this is fixed? or assigning to me if it still needs more work? thank you

[overlookmotel]

No, #11632 fixed this case:

for (const x in {}) var y = 1;
//                  ^^^^^ error

This issue is about a different case:

for (var x in {}) {}
//   ^^^^^ error?

The question is: Should this 2nd case be an error or not?

ESLint makes it an error (ESLint playground), so I'd assume it should be. But we have code that specifically looks for this pattern and makes it not an error, so at someone decided that's how it should be at some point. I'm not sure why.

If we decide that it should be an error, I'll be happy to fix it. I just don't know the linter well, so I thought I'd check first.

[camc314]

yes, it should be an error imo

 node
Welcome to Node.js v22.15.0.
Type ".help" for more information.
> for (var k = 123;false;) {}
undefined
> k
123
> 

It doesn't make sense that you can access k after the for loop scope has exited

[camc314]

To add, this behaviour has been there since the rule was originally added. I don't think it was deliberate are there were no tests for this behaviour.

1aa1129