fix: validate parent node mutations

[ShiroKSH]

Summary

• Reject parent-node mutations whose reference or replaced node belongs to another parent.
• Reject inserting an ancestor into a descendant, preventing cyclic trees.
• Preserve the existing detached insertBefore(node, node) no-op.

Tests

• npm test
• Worker bundle repro for an invalid foreign reference

[WebReflection]

a bit defensive which is a bit of a slippery slope or not the goal of this project or goodbye performance but it seems fair. may I ask what's the use case for this, beside stricter adherence with DOM standard which is not a goal of this project?

[ShiroKSH]

Fair point. I do not have a production use case.

I found it while testing invalid mutation paths. The failure is not just a spec mismatch. A foreign reference can put a node in one parent list while its parentNode points to another. replaceChild can remove a node from the wrong parent. An ancestor insertion can create a cycle and make traversal hang.

If invalid input is out of scope then this is not worth the hot path cost. Happy to close the PR.

[WebReflection]

my time is limited and logic in the hot path for something intentionally broken is probably not worth a fix, imho. Closing but thanks for expanding on that.