Daily Digest

Loading brief…

• Lower conditional merges to predecessor copies

  github:NVIDIA/warp7/13/2026crashes-stabilitywarpcudacodegencontrol-flowundefined-behaviorillegal-memory-access

  Warp lowering of conditional merges can leave operands uninitialized along some paths, and a nested case leaked a branch value into its sibling branch. Under register pressure this surfaced as illegal CUDA memory access.

  ⚡ Warp codegen correctness under control-flow is foundational—any undefined behavior undermines GPU safety for simulation and learning workloads.

• [Bug Report] Duplicate imported classes

  github:isaac-sim/IsaacLab7/14/2026tooling-dxisaac-labpython-importsimulationappreproducibilityconfigstype-system

  Depending on SimulationApp startup/import order, Isaac Lab classes can be imported twice and end up with mismatched parent classes. This breaks type checks (e.g., config objects failing isinstance) and can cause subtle downstream failures.

  ⚡ Import/initialization determinism in Isaac Lab needs hard guarantees to keep large-scale training and evaluation pipelines reproducible.

• MjSpec.to_xml changes joint/state-vector ordering after attach round-trip

  github:google-deepmind/mujoco7/14/2026integrationmujocomjcfmodel-editingroundtripstate-vectorjoint-ordering

  MuJoCo's MjSpec.to_xml() can change joint/state-vector ordering after exporting and re-importing modular MJCF using asset/model and attach. This breaks assumptions about stable state ordering across a round-trip.

  ⚡ Round-trip integrity across modular model composition is a prerequisite for trustworthy sim-based validation and long-lived robot assets.

• Handle empty mesh decomposition results as failures

  github:newton-physics/newton7/14/2026asset-pipelinenewtonasset-importconvex-decompositionvhacdcoacderror-handling

  approximate_meshes() silently treats an empty CoACD/V-HACD decomposition as success and leaves meshes unchanged. It neither warns nor raises even when raise_on_failure=true, violating the documented fallback/contract.

  ⚡ Robust convex decomposition with explicit failure modes is critical for scalable asset ingestion and reliable physics behavior.

• Investigate VBD cloth quality and performance gap in teleoperation

  github:newton-physics/newton7/14/2026crashes-stabilitynewtonteleoperationclothvbdphysxprofiling

  A Newton teleoperation workload with two-way coupled MJWarp robot and VBD cloth runs ~10+ FPS versus 18–20 FPS on PhysX after initial optimization. The hypothesis is VBD needs more substeps for comparable cloth quality, requiring profiling to isolate physics c

  ⚡ Teleop workloads expose real-time physics/performance gaps that will directly impact interactive robot development and validation loops.