Issue with deleteNode when deleting all nodes

[marcelino-pensa]

I have implemented a procedure for a fading memory octomap, meaning that it will forget things that it saw very long ago by deleting those nodes using deleteNode.
If I end up deleting all nodes from the octree, I noticed that the root node remains alive, and that the octree is saying that my whole volume is occupied.
After I delete all nodes from the octree, only one node remains, and that one is the one whose size is equal to tree.getResolution() * pow(2.0, 16.0) (i.e., the root). This node cannot be deleted, as deleteNodeRecurs does not delete a node if node != this->root.

My questions are:

• Is there any reason why my octomap reverts back to the root as I delete all nodes? Is this the expected behavior? Why not deleting the root as well? Maybe it can't delete the root because deleteNodeRecurs does not delete the root...
• Why can't we delete the root in deleteNodeRecurs?

[marcelino-pensa]

If my question isn't clear, this is what I mean by deleting all nodes:

    for (octomap::OcTree::leaf_iterator it = tree_.begin_leafs(),
                                       end = tree_.end_leafs();
                                       it != end; ++it) {
        octomap::OcTreeKey key = it.getKey();
        tree_.deleteNode(key, it.getDepth());
    }

If we delete all nodes using the method above, there will be one remaining node in the tree that is impossible to delete (the root)

[ahornung]

Yes, this seems like an omission that should be fixed. Looking at the implementation of clear, there's no reason not to delete the root as well.

[ahornung]

I haven't tested it yet, but PR #288 should fix this as well. Does that work for you?

[ahornung]

Looking back at the changelog, this issue was fixed long ago with #11 (v1.6.0) and it's probably a regression.

[marcelino-pensa]

I am running on v1.9.0, and the problem still exists

[xiaofengguorui]

I have the same issu, and when I delete all nodes, I try to updateNode failed.