Fix RuntimeError when click handler deletes the clicked element during testing

[falkoschindler]

Motivation

Fixes #5937. When a button's click handler clears its parent container (destroying the button itself), User.click() raises RuntimeError: dictionary changed size during iteration because element._event_listeners is mutated mid-loop via _handle_delete() -> _event_listeners.clear().

Implementation

After each handler fires, check element.is_deleted and break out of the listener loop. This avoids the dict-mutation error without copying the dict, and — more importantly — matches real browser behavior: in production, element._handle_event does a direct dict lookup per listener_id (no iteration), so once an element is deleted, subsequent browser messages simply never arrive and no more listeners fire. The is_deleted + break approach faithfully mirrors this, preventing false-positive tests where handlers would run on already-deleted elements.

The test registers three click handlers on a button inside the container it clears — the first handler fires, the second (rebuild) deletes and recreates the button, and the third must not fire since the element is now deleted.

Progress

• The PR title is a short phrase starting with a verb like "Add ...", "Fix ...", "Update ...", "Remove ...", etc.
• The implementation is complete.
• This PR does not address a security issue.
• Pytests have been added.
• Documentation is not necessary.
• No breaking changes to the public API.

[evnchn]

Hi — this is Claude Code, acting on behalf of @evnchn.

The core question isn't just "does it stop erroring?" but "does the test simulation still behave like real life?"

The list() snapshot fix prevents the RuntimeError, but it subtly diverges from production behavior. An alternative: check element.is_deleted after each handler and break — no copy, and it matches what actually happens in the real dispatch path.

Detailed analysis

Real event dispatch (element._handle_event): the browser sends one message per listener_id. The server does a direct dict lookup — no iteration. If a handler deletes the element, subsequent messages from the browser simply target a nonexistent element and never fire.

list() snapshot approach: prevents the RuntimeError by iterating a copy, but if a handler deletes the element mid-loop, the remaining listeners in the snapshot still fire — on an already-deleted element. This can't happen in real life.

is_deleted + break approach: iterates the dict directly (no copy, O(1) memory). After each handler fires, checks element.is_deleted — if the element was destroyed, breaks out of the loop. This mirrors real behavior: once an element is deleted, no more listeners fire.

The difference only matters for multi-listener scenarios, but aligning the simulation with reality prevents a class of false-positive tests where handlers run on deleted elements.

Commit: 7445dee on this branch.

[evnchn]

Feel free to merge if you agree, or discuss if not. Thanks!

[evnchn]

And be sure to update PR desc.

[falkoschindler]

Great catch, @evnchn! I didn't think about checking the real-world behavior, but you're completely right. I updated the pytest and PR description accordingly. 👍🏻

[evnchn]

The updated test, comment, and PR description all show that me, you, test suite, and reality, all align 4-way.