into_new_tree() Panics if Node has no Children

[vixi-void]

When NodeMut::into_new_tree() is called on a node that has zero children it panics in NodeMut::try_append_subtree_as_child() during an expect("tree is not empty"). My understanding is that the intention of this expect() call, is to state that at this point, the tree being non-empty is intended to be an invariant. If this is intended to be an invariant, it has been violated. Otherwise, if this is intended behavior, I think that the documentation for NodeMut::into_new_tree() should state that it panics when called on a node with zero children. Here is a minimal example

    use orx_tree::*;

    let mut tree = DynTree::new(0);
    tree.root_mut().push_child(1);
    // Does not panic if we add a children to 1
    // tree.root_mut().children_mut().next().unwrap().push_child(2);
    let tree2: DynTree<_> = tree
        .root_mut()
        .children_mut()
        .nth(0)
        .unwrap()
        .into_new_tree();

I don't have much experience creating issues, so if there's anything else I should have included please let me know. Thank you, I've been enjoying your crate.

[vixi-void]

The issue peaked my interest again and I found where the invariant is broken. When NodeMut::into_new_tree() is called on a node with zero children, within multiple nested calls, we eventually get to

orx-tree/src/common_traits/from_depth_first_iter.rs

Lines 212 to 233 in 18d4afc

	fn try_from(value: DepthFirstSequence<V::Item, I>) -> Result<Self, Self::Error> {
	let mut iter = value.0.into_iter();
	match iter.next() {
	None => Ok(Tree::default()),
	Some((d, root)) => match d {
	0 => {
	let mut tree = Tree::new_with_root(root);
	match tree.root_mut().try_append_subtree_as_child(iter, 0) {
	Ok(_) => Ok(tree),
	Err((depth, succeeding_depth)) => {
	Err(DepthFirstSequenceError::DepthIncreaseGreaterThanOne {
	depth,
	succeeding_depth,
	})
	}
	}
	}
	_ => Err(DepthFirstSequenceError::NonZeroRootDepth),
	},
	}
	}
	}

where the argument value is a depth first search sequence iterating starting at the root of said node with zero children. Then, in this function, when you get to

orx-tree/src/common_traits/from_depth_first_iter.rs

Line 219 in 18d4afc

match tree.root_mut().try_append_subtree_as_child(iter, 0) {

it is assumed that iter is non-empty and note that iter being non-empty is a precondition for NodeMut::try_append_subtree_as_child(). Yet just previously

orx-tree/src/common_traits/from_depth_first_iter.rs

Line 214 in 18d4afc

match iter.next() {

the singular element of iter (the root) was consumed and thus iter is empty, violating the precondition.

[orxfun]

@vixi-void . Thank you very much for finding the problem and creating such a detailed issue. This is super helpful, I am working on it.

[orxfun]

@vixi-void feel free to take a look at the PR #184 if you have time.

[vixi-void]

I'm really glad you appreciated it. Thank you for working on it! I took a look at the issue and your commits. Everything looks good to me. I checked out the branch and ran gdb and I didn't see any issues either. Should this issue be closed when you merge the branch into main or should I have closed it with this comment?

[orxfun]

Thanks for checking. I'd say it's better to have it closed when the PR is merged.