Fix pre-existing thread-caps test failure: issue779_layer_sweep.py missing load_dotenv-before-torch
Highlight text on any card (body, plan, or an event) to anchor a comment, or leave a whole-task comment here. Mention @claude to summon a reply.
No comments yet.
Overview / Motivation
Surfaced by task #876's implementer (2026-07-02): tests/test_shared_vm_thread_caps.py::test_no_new_torch_before_dotenv_vm_entrypoints FAILS pre-existing on pristine main — scripts/issue779_layer_sweep.py (commit fd1d5512e8, task #779) imports torch/numpy at module top with no orchestrate.env.load_dotenv() call before them and no grandfather entry. Every task whose Step 9c touched-file subset includes this workflow-invariant test inherits the red.
Goal
Make test_no_new_torch_before_dotenv_vm_entrypoints pass on main: add orchestrate.env.load_dotenv() before the torch import in scripts/issue779_layer_sweep.py (preferred — the shared-VM thread-cap rule requires load_dotenv before torch import so OMP caps freeze correctly), OR add a grandfathered allowlist entry with an inline reason (the script self-caps via torch.set_num_threads at ~:486).
Constraints
- One-liner class; do not change the script's behavior otherwise.
- Coordinate with the live #779 session (task at followups_running) — the file is on
main; its worktree inherits on rebase. uv run pytest tests/test_shared_vm_thread_caps.pypasses after the fix; ruff clean.
Provenance
- Surfaced by: #876 implementer round 1 (pre-existing-on-main test failure, 0 introduced by the #876 diff).