Issue with ports not being remapped with subtreeplus on v3.8

[Rid0n]

Here's my code.

I have the outermost tree passing down values to a subtree which is then passing down values to another subtree which ticks a navigation tree. Everything with the __autoremap flag set to true. Very simple.

<root main_tree_to_execute="Tree1">
<include path="./navigation_tree.xml"/>

<BehaviorTree ID="Tree1">

    <Sequence>

        <SubTreePlus ID="Tree2" __autoremap="true"/>

    </Sequence>

</BehaviorTree>

<BehaviorTree ID="Tree2">

    <Sequence>

        <SubTreePlus ID="Tree3" __autoremap="true"/>

    </Sequence>

</BehaviorTree>

<BehaviorTree ID="Tree3">

    <Sequence>

        <SubTreePlus ID="NavigateTree" __autoremap="true"/>

    </Sequence>

</BehaviorTree>

</root>

I expect there not to be any issues, however for some reason I'm getting an error

[bt_navigator]: Exception when loading BT: Blackboard::get() error. Missing key [node]

I found that the issue lies specifically in the switch from Tree2 to Tree3. Every other subtree node does successfully remap everything.

If I change one line
<SubTreePlus ID="Tree3" __autoremap="true"/>
To
<SubTreePlus ID="Tree3" goal="{goal}" node="{node}" bt_loop_duration="{bt_loop_duration}" tf_buffer="{tf_buffer}" server_timeout="{server_timeout}" />
It all works fine.

What's up here?

[facontidavide]

Version 4.X has been in development for almost 1 year now. Keep maintaining 3.8 is really a struggle for me.

Said that, would you be so kind to give me a unit test that I can use to reproduce the issue?

[facontidavide]

These are my current tests: https://github.com/BehaviorTree/BehaviorTree.CPP/blob/v3.8/tests/gtest_subtree.cpp

[facontidavide]

I was able to reproduce here: https://github.com/BehaviorTree/BehaviorTree.CPP/tree/issue563
it is indeed a bug. I will let you know once it is is fixed

[facontidavide]

Hi, problem fixed in branch issue563.

Would you be so kind to test it?

Once you confirm, I will merge and release a new version

[Rid0n]

still there for me D:
it also broke the delicate ways in which it did work haha

trying to find a test case where that'll show

[facontidavide]

oh god, don't tell me that!
Please tell me more, this issues is very important and I want to get it fixed as soon as possible.

[facontidavide]

also, if you experienced a regression, please share that with me. I hope I am not breaking anything with these changes (my unit tests pass)

[Rid0n]

So, the unit test passes fine, but when I'm trying a simple thing with the nav2 stack, it fails.

<root main_tree_to_execute="Tree">

<BehaviorTree ID="Tree">

    <Sequence>

        <SubTreePlus ID="Main" __autoremap="true"/>

    </Sequence>

</BehaviorTree>

<BehaviorTree ID="Main">

    <Wait wait_duration="1"/>

</BehaviorTree>

</root>

fails to instantiate tree with
Blackboard::get() error. Missing key [node]

the bt_navigator node crashes entirely, so it looks like there's something nasty together with nav2 specifically.
a bunch of nav2_behavior_tree unit tests now fail too