DiffOpMove does not work in debug mode (only works in release mode)

[Gronis]

While I was exploring the possibility of minimising the binary size of (my version of) the js-framework-benchmark for leptos framework, I ran into a problem where the "swap rows" button didn't work in debug mode. It turns out it only works in release mode. 😬

I have now found the issue which seems to be related to the function EachItem.get_opening_node which work differently in release and debug mode. In debug mode, the opening node is always a Comment type node. In release mode it's a normal Element type node. In debug mode, later when mount_child(MountKind::Before(&opening), &each_item); is called to insert the nodes at the right positions, it does not work because opening is a comment node. Changing the behavior of get_opening_node to work the same as release in mode fixes this problem.

Because I don't know the intent of having different implementations of get_opening_node, I decided not to open a pull request of my "fix". However, I still think this issue needs to be resolved, since a broken DiffOpMove implementation in debug builds are unacceptable.

This is the change I did to "fix" the issue:

leptos_dom/src/components.rs

impl Mountable for ComponentRepr {

    // ...

    fn get_opening_node(&self) -> web_sys::Node {
        // #[cfg(debug_assertions)]
        // return self._opening.node.clone();

        // #[cfg(not(debug_assertions))]
        return if let Some(child) = self.children.get(0) {
            child.get_opening_node()
        } else {
            self.closing.node.clone()
        };
    }

    // ...

}

leptos_dom/src/components/each.rs

impl Mountable for EachItem {

    // ...

    #[inline(always)]
    fn get_opening_node(&self) -> web_sys::Node {
        // #[cfg(debug_assertions)]
        // return self.opening.node.clone();

        // #[cfg(not(debug_assertions))]
        return self.child.get_opening_node();
    }

    // ...

}

[gbj]

Thanks. The <For/> component is toward the tail end of a significant rewrite by @jquesada2016 (see #533, #897), which will likely resolve this issue in any case.

In the meantime, if you have a solution please do open a PR so we can test it.

Edit: Had my brain off for a second, I see your solution included in the issue here and have tested it and it does seem to work here. I'll test it against some other cases. Thanks.