I have verified that the issue does not occur with Node 18 and yet it occurs systematically with Node 20 and up.

@chharvey Please review #53431

it seems like it has somthing to do with x being an array saved in constant. example:

const x = ["x"];

const a = new Set([x, ["y"]]);
const b = new Set([x, ["y"]]);

assert.deepStrictEqual(a, b); // AssertionError

but:

const a = new Set([["x"], ["y"]]);
const b = new Set([["x"], ["y"]]);

assert.deepStrictEqual(a, b); // null

can i work on this?

can i work on this?

Yes!!! Have a look at #53431

@DanielBelz1997 Yup, your analysis is correct. Take a look at the setEquiv function here.

Starting on line 447, a new "SafeSet" is created that contains all the values in a that are not in b (as determined by b.has(), whch checks by reference). When you create a reference const x = ['x'], it doesn’t get put into that SafeSet since it’s in both sets. Whereas a newly constructed object would be put in since it can’t be in both sets.

The bug lies on line 468, where we loop over the values in b. Since the reference x is in b but not in the SafeSet, the call on line 471, setHasEqualElement(‹SafeSet›, ‹x›, ...), fails. In your second example you have no references, so all the values in a get put into that SafeSet when it’s created. Then when we loop over b, the call setHasEqualElement(...) passes for all those values.

@chharvey thank you for the direction! I will get on that as soon as possible.

This is a great catch! The bug seems to only occur for sets where both contain a reference identical object and another non-reference identical object.

To fix it, it should be enough to add a check here to verify the reference identical one does not exist in the other set

        if (!a.has(val) && !setHasEqualElement(set, val, strict, memo))

An alternative would be to undo some of the set changes from #46593. It was not an issue before due to always adding all objects to the intermediate set.

I would just run the benchmark on both options and see what has less impact.

@BridgeAR I have added your simple fix to #53431