Unit tests fail in Python 3.11 (v0.62 and main: c09d949)

[gavincangan]

Please provide the below information in addition to your issue:

1. cuRobo installation mode: python
2. python version: 3.11.7
3. Isaac Sim version (if using): N/A

---

Issue Details

I ran the unit tests after pip installing cuRobo and they error out. Here's the output:

gavin@omen:~/repos/curobo$ python3 -m pytest .
================== test session starts ====================
platform linux -- Python 3.11.7, pytest-7.4.4, pluggy-1.3.0
rootdir: /home/gavin/repos/curobo
configfile: pyproject.toml
plugins: anyio-4.2.0, torchtyping-0.1.4, typeguard-4.1.5
collected 3 items / 27 errors / 1 skipped

========================== ERRORS ===================
___________________________________________________________________________________________________________ ERROR collecting tests/cost_test.py ____________________________________________________________________________________________________________
tests/cost_test.py:15: in <module>
    from curobo.geom.sdf.world import WorldCollisionConfig, WorldPrimitiveCollision
src/curobo/geom/sdf/world.py:22: in <module>
    from curobo.geom.types import Cuboid, Mesh, Obstacle, WorldConfig, batch_tensor_cube
src/curobo/geom/types.py:36: in <module>
    @dataclass
../../miniforge3/envs/torch11.8/lib/python3.11/dataclasses.py:1230: in dataclass
    return wrap(cls)
../../miniforge3/envs/torch11.8/lib/python3.11/dataclasses.py:1220: in wrap
    return _process_class(cls, init, repr, eq, order, unsafe_hash,
../../miniforge3/envs/torch11.8/lib/python3.11/dataclasses.py:958: in _process_class
    cls_fields.append(_get_field(cls, name, type, kw_only))
../../miniforge3/envs/torch11.8/lib/python3.11/dataclasses.py:815: in _get_field
    raise ValueError(f'mutable default {type(f.default)} for field '
E   ValueError: mutable default <class 'curobo.geom.types.Material'> for field material is not allowed: use default_factory

I tried some hacky fixes:

# in src/curobo/geom/types.py
-    material: Material = Material()
+    material: Material = field(default_factory=Material)

# in src/curobo/rollout/rollout_base.py
-    goal_pose: Pose = Pose()
+    goal_pose: Pose = field(default_factory=Pose)

# in src/curobo/wrap/reacher/trajopt.py
-    traj_evaluator_config: TrajEvaluatorConfig = TrajEvaluatorConfig()
+    traj_evaluator_config: TrajEvaluatorConfig = field(default_factory=TrajEvaluatorConfig)

But now, I have this warp error

E   warp.codegen.WarpCodegenError: Error while parsing function "compute_matrix_to_quat" at /home/gavin/repos/curobo/src/curobo/geom/transform.py:401:
E       out_v[0] = wp.index(out_q, 3)
E   ;Couldn't find function overload for 'index' that matched inputs with types: [warp.types.quatf, int32]

[balakumar-s]

A breaking API change happened in warp. We are pushing a fix today to cuRobo. In the meantime, can you pip install warp-lang==0.9.0?

[gavincangan]

Thanks for the prompt response! So I did pip install warp-lang==0.9.0, but needed to make a few other changes for the unit tests to run: https://github.com/gavincangan/curobo/tree/fix/unit_tests

[balakumar-s]

The library has not been tested on python>3.10.

Regarding your change to torch.Size(https://github.com/gavincangan/curobo/commit/ff64469cdb33490a458b1ced921f48a6235624f8) What pytorch version are you using?

The issue with wp.index is fixed in the latest commit, so newer warp should also work.

[gavincangan]

The library has not been tested on python>3.10.

Regarding your change to torch.Size(gavincangan@ff64469) What pytorch version are you using?

I'm using v2.1.2

The issue with wp.index is fixed in the latest commit, so newer warp should also work.

Ah, thank you, I'll switch to the newer version then :)