rmf_task appears to use rmf_traffic directly without declaring it in package.xml

Summary

While reviewing ROS 2 package dependency metadata, I noticed that rmf_task appears to use rmf_traffic directly, but rmf_traffic does not seem to be declared in rmf_task/package.xml.

This may make the package rely on rmf_battery or another intermediate package to provide rmf_traffic transitively.

Evidence

Direct usage

rmf_task directly includes rmf_traffic headers in production headers:

• include/rmf_task/Phase.hpp:24: #include <rmf_traffic/Time.hpp>
• include/rmf_task/Log.hpp:27: #include <rmf_traffic/Time.hpp>
• include/rmf_task/Header.hpp:21: #include <rmf_traffic/Time.hpp>
• include/rmf_task/Header.hpp:22: #include <rmf_traffic/agv/Graph.hpp>
• include/rmf_task/Estimate.hpp:26: #include <rmf_traffic/Time.hpp>
• include/rmf_task/Estimate.hpp:27: #include <rmf_traffic/agv/Planner.hpp>
• include/rmf_task/State.hpp:26: #include <rmf_traffic/agv/Planner.hpp>

The manual source review found extensive use of rmf_traffic::Time, rmf_traffic::Duration, rmf_traffic::Trajectory, rmf_traffic::agv::Planner, and rmf_traffic::agv::Plan::Start.

Current package.xml

I could not find a direct declaration of rmf_traffic in rmf_task/package.xml. The package currently declares:

<depend>rmf_battery</depend>
<depend>rmf_utils</depend>

Observed during build/test

During build/test tracing, files from rmf_traffic were accessed, for example:

• lib/rmf_traffic/cmake/rmf_traffic-config.cmake
• lib/rmf_traffic/cmake/rmf_traffic-targets.cmake
• share/colcon-core/packages/rmf_traffic
• lib/librmf_traffic.so

Transitive path

The dependency appears to be reachable transitively through:

rmf_task -> rmf_battery -> rmf_traffic

Suggested fix

If these traffic planning types are direct dependencies of rmf_task, would it make sense to add:

<depend>rmf_traffic</depend>

Notes

This issue is not claiming that the package currently fails to build. The concern is that the package directly uses rmf_traffic while relying on a transitive dependency path to make it available. The observation is based on package metadata, source usage, dependency closure analysis, and build/test file-access tracing.

Could you please confirm whether this dependency is intentionally left implicit through the transitive dependency path shown above, or whether adding an explicit dependency to package.xml would be appropriate?

I would be happy to open a small PR adding the dependency if that matches the intended package metadata.

Start with rmf_task/package.xml and the listed production headers, especially Phase.hpp, Header.hpp, Estimate.hpp, and State.hpp. Confirm that the headers use rmf_traffic directly while only rmf_battery and rmf_utils are declared. Done means the metadata reflects the confirmed direct dependency and the package build/tests still pass.