Split Direct MARL space and lifecycle tests

[nvsekkin]

Summary

• Reuse the config-only make_empty_direct_marl_env_cfg fixture in the simulator-backed Direct MARL lifecycle test.
• Add a kitless CPU unit test for agent counts and Gymnasium space construction.
• Retain CPU/CUDA initialization, constructor-to-space wiring, and SimulationContext cleanup coverage.

This is PR 4 of the focused replacement series for draft #6526.

Type of change

• Test refactor (no production behavior change)

Checklist

• I have read and understood the contribution guidelines
• I have run the pre-commit checks with ./isaaclab.sh --format
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• I have updated the changelog and the corresponding version in the extension's config/extension.toml file
• I have added my name to the CONTRIBUTORS.md or my name already exists there

[greptile-apps]

Greptile Summary

This PR splits the single test_direct_marl_env.py integration test into two: a simulator-backed lifecycle test (refactored) and a new kitless unit test (test_direct_marl_env_spaces.py) that exercises agent-count and Gymnasium-space construction without launching Isaac Sim.

• test_direct_marl_env.py: Replaces the inline helper with the shared make_empty_direct_marl_env_cfg fixture and adds post-close assertions, but moves the DirectMARLEnv constructor call outside the try/finally block.
• test_direct_marl_env_spaces.py: Creates a DirectMARLEnv via object.__new__, stubs scene and sim with SimpleNamespace, and calls _configure_env_spaces() directly to validate space shapes without starting the simulator.

Confidence Score: 3/5

The refactored lifecycle test has a gap: if DirectMARLEnv construction throws, close() is never called and SimulationContext leaks into the next parametrized test case.

The constructor call in test_direct_marl_env.py sits outside the try/finally block, so an init-time failure skips cleanup and can taint the subsequent CPU parametrize run. The kitless test in test_direct_marl_env_spaces.py calls a private method on a hand-assembled object, which is fragile if _configure_env_spaces reads any additional attribute initialised by init.

test_direct_marl_env.py — constructor/try/finally ordering; test_direct_marl_env_spaces.py — object.new + private-method approach.

Important Files Changed

Filename	Overview
source/isaaclab/test/envs/test_direct_marl_env.py	Refactors lifecycle test to use shared fixture and adds post-close assertions; constructor call is outside try/finally, risking a leaked SimulationContext if init raises.
source/isaaclab/test/envs/test_direct_marl_env_spaces.py	New kitless unit test for agent counts and space construction; calls a private method on a manually-assembled object, which is fragile to internal refactors.
source/isaaclab/changelog.d/core-test-direct-marl.skip	Changelog skip fragment for a test-only change; no functional concerns.

Sequence Diagram

%%{init: {'theme': 'neutral'}}%%
sequenceDiagram
    participant Test as test_initialization_and_close
    participant SU as sim_utils
    participant Env as DirectMARLEnv

    Test->>SU: create_new_stage()
    Test->>Env: "DirectMARLEnv(cfg=...) [OUTSIDE try/finally]"
    Note over Env: __init__ sets up SimulationContext,<br/>scene, spaces via _configure_env_spaces()
    Test->>Test: try block starts
    Test->>Env: assert not _is_closed
    Test->>SU: assert SimulationContext.instance() is env.sim
    Test->>Env: assert num_agents, spaces, state_space
    Test->>Test: finally block
    Test->>Env: env.close()
    Note over Env: clear_instance(), _is_closed = True
    Test->>Test: assert env._is_closed
    Test->>SU: assert SimulationContext.instance() is None

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
sequenceDiagram
    participant Test as test_initialization_and_close
    participant SU as sim_utils
    participant Env as DirectMARLEnv

    Test->>SU: create_new_stage()
    Test->>Env: "DirectMARLEnv(cfg=...) [OUTSIDE try/finally]"
    Note over Env: __init__ sets up SimulationContext,<br/>scene, spaces via _configure_env_spaces()
    Test->>Test: try block starts
    Test->>Env: assert not _is_closed
    Test->>SU: assert SimulationContext.instance() is env.sim
    Test->>Env: assert num_agents, spaces, state_space
    Test->>Test: finally block
    Test->>Env: env.close()
    Note over Env: clear_instance(), _is_closed = True
    Test->>Test: assert env._is_closed
    Test->>SU: assert SimulationContext.instance() is None

Loading

_{Reviews (1): Last reviewed commit: "test: Split Direct MARL space coverage" | Re-trigger Greptile}

[mataylor-nvidia]

I have added a comment ontop of greptiles suggestion