1821790 - document.adoptNode (and others causing it) changes the prototype of the adopted node

Status: RESOLVED DUPLICATE of bug 1470017

(Core :: DOM: Core & HTML, defect)

Description

[Ksenia Berezina [:ksenia]]

Attachment: 118350.html

We've received a report in https://github.com/webcompat/web-bugs/issues/118350 about the discrepancy between Firefox and Chrome when calling instanceof HTMLElement for an element within an iframe.

The reporter provided a testcase and mentioned that this is causing issues for them in production since Firefox 109.

It returns false in Firefox when instanceof HTMLElement is called. And returns true when the iframe window (instanceof iframe.contentWindow.HTMLElement) checked instead. The results in Chrome are exact opposite.

mozregression points to https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=5199ec2d73fa7daf423b72416be41383ac1042dc&tochange=95b6c826ad974a967364cb1f137bd8a766051d0a, so I assume this is related to bug1360715

Comment 1

[Ksenia Berezina [:ksenia]]

Setting dom.webidl.crosscontext_hasinstance.enabled to true makes their testcase pass and Firefox starts matching Chrome, but only for instanceof HTMLElement, which returns true. instanceof iframe.contentWindow.HTMLElement is always false in Chrome. Wonder which behaviour is correct here?

Hi Kagami, could you please take a look at this?

Comment 2

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1360715

Comment 3

[Kagami Rosylight [:saschanaz] (they/them)]

This is not a regression from bug 1360715, this is more of the node adoption behavior rather than of the instanceof behavior.

el2 instanceof HTMLElement is true on Gecko too before el.appendChild(e2), but the adoption somehow magically changes el2's prototype, which is very surprising.

Comment 4

[Kagami Rosylight [:saschanaz] (they/them)]

Attachment: bug1821790.html

Replaced instanceof with Object.getPrototypeOf(), and turns out Gecko has been doing this forever (as version 25 was doing the same thing).

Comment 5

[Kagami Rosylight [:saschanaz] (they/them)]

Seems it's known:

Comment 6

[Kagami Rosylight [:saschanaz] (they/them)]

I moved the webcompat report to the bug, hope it works better!

Comment 7

[Ksenia Berezina [:ksenia]]

That's good to know, thanks for investigating!

Comment 8

[Andrew McCreight [:mccr8]]

It sounds like this is old behavior and not actually a regression.