Btw... By investigating on this bug, I realized that how the Blackboard::set itself is handled is pretty weird wrt port remappings. Take this tree for instance....

<?xml version="1.0"?>
<root BTCPP_format="4" main_tree_to_execute="BehaviorTree">
    <!-- ////////// -->
    <BehaviorTree ID="BehaviorTree">
        <Sequence>
            <SetBlackboard output_key="pos" value="0.0;0.0" />
            <KeepRunningUntilFailure>
                <ForceSuccess>
                    <Sequence>
                        <UpdatePosition pos_in="{pos}" x="0.1" y="0.2" pos_out="{pos}"/>
                        <SubTree ID="A" _autoremap="true" new_pos="2.2;2.4" />
                        <Sleep msec="500"/>
                    </Sequence>
                </ForceSuccess>
            </KeepRunningUntilFailure>
        </Sequence>
    </BehaviorTree>
    <BehaviorTree ID="A">
        <Sequence>
            <SetBlackboard output_key="pos" value="3.0;5.0" />
        </Sequence>
    </BehaviorTree>
    <!-- ////////// -->
    <TreeNodesModel>
        <SubTree ID="BehaviorTree" description=""/>
        <SubTree ID="A" description="">
            <input_port default="1.0;1.0" name="new_pos">Updated position</input_port>
        </SubTree>
    </TreeNodesModel>
    <!-- ////////// -->
</root>

It's completely useless... I grant you that, but it should work... Some key points to understand it:

• The KeepRunningUntilFailure and ForceSuccess are just a combo trick to repeat the sequence... You can repeat it two times to have the same result
• pos in the root blackboard is strongly typed (i.e. custom type "Position2D") because of UpdatePosition node
• It is auto-remapped in the subtree A where I use an on the fly instantiated entry to push into pos essentially an Any of type string (the type checking to resolve the port remappings are not triggered because of the fact that the string value is not strongly typed)

The first iteration runs without any issue, but on the second I would have a Any::copyInto failure due to the fact that the UpdatePosition is trying to push there through the SetOutput call an object of type Position2D, but encounters a string....

Please note that this error is not due whatsover to my modifications proposed in the PR and it's not resolved by them either... Maybe we shall rethink a little bit how the current port remapping and type checking work (for instance... the type check when there is an autoremap to resolve is not as flexible as when there is not: in the latter case string conversion or math acceptable casts are accepted, while in the former not....)

I hope the explanation is sufficiently clear.. @facontidavide what do you think about it ?

very good work. thanks for the tests!

about the issue you mentioned, I will need to think about it