configure ruff - #260
Conversation
Signed-off-by: Jinzhe Zeng <jinzhe.zeng@rutgers.edu>
for more information, see https://pre-commit.ci
Signed-off-by: Jinzhe Zeng <jinzhe.zeng@rutgers.edu>
Signed-off-by: Jinzhe Zeng <jinzhe.zeng@rutgers.edu>
Signed-off-by: Jinzhe Zeng <jinzhe.zeng@rutgers.edu>
Signed-off-by: Jinzhe Zeng <jinzhe.zeng@rutgers.edu>
Why is the type of |
Signed-off-by: Jinzhe Zeng <jinzhe.zeng@rutgers.edu>
Signed-off-by: Jinzhe Zeng <jinzhe.zeng@rutgers.edu>
| "exploration_scheduler" | ||
| ].value_from_expression = if_expression( | ||
| _if=(scheduler_step.outputs.parameters["converged"] == True), | ||
| _if=(scheduler_step.outputs.parameters["converged"] is True), |
There was a problem hiding this comment.
P1 — Preserve dflow’s overloaded equality expression. This value is a dflow OutputParameter, not a Python boolean. is True evaluates immediately to the literal Python value False, so all three if_expression calls always select the recursive _else branch. When the scheduler has converged, that branch references next_step even though its when=... == false condition skipped it, leaving workflow outputs unresolved. Restore the dflow comparison at lines 446, 451, and 456 and suppress E712 locally; the comparison is expression construction.
| _if=(scheduler_step.outputs.parameters["converged"] is True), | |
| _if=(scheduler_step.outputs.parameters["converged"] == True), # noqa: E712 |
Codex quota is about to reset, so I am using the remaining token budget to review this PR now.
Coding agent: Codex
Codex version: codex-cli 0.144.6
Model: gpt-5.6-sol
Reasoning effort: xhigh
| __all__ = [ | ||
| "ConfFilter", | ||
| "ConfFilters", | ||
| "ConfSelector", | ||
| "ConfSelectorFrames", | ||
| "conf_filter_styles", | ||
| ] |
There was a problem hiding this comment.
P2 — Keep the concrete filter classes in wildcard exports. Adding __all__ changes the package API, but this list omits three classes that are still imported above. Before this PR, from dpgen2.exploration.selector import * exposed them; afterward they silently disappear.
| __all__ = [ | |
| "ConfFilter", | |
| "ConfFilters", | |
| "ConfSelector", | |
| "ConfSelectorFrames", | |
| "conf_filter_styles", | |
| ] | |
| __all__ = [ | |
| "BoxLengthFilter", | |
| "BoxSkewnessConfFilter", | |
| "ConfFilter", | |
| "ConfFilters", | |
| "ConfSelector", | |
| "ConfSelectorFrames", | |
| "DistanceConfFilter", | |
| "conf_filter_styles", | |
| ] |
Codex quota is about to reset, so I am using the remaining token budget to review this PR now.
Coding agent: Codex
Codex version: codex-cli 0.144.6
Model: gpt-5.6-sol
Reasoning effort: xhigh
|
|
||
|
|
||
| def _sample_sphere(): | ||
| rng = np.random.default_rng() |
There was a problem hiding this comment.
P2 — Preserve a seedable PKA direction. Constructing a fresh entropy-seeded generator inside _sample_sphere() disconnects this operation from np.random.seed(), and neither this helper nor make_lmp_input() accepts a seed/generator. Existing callers therefore lose their only way to reproduce the generated direction. Thread an explicit generator or seed through the API, or retain the legacy global RNG with a targeted # noqa: NPY002; add a deterministic regression test. A complete fix requires an API decision, so a local suggestion is not safe.
Codex quota is about to reset, so I am using the remaining token budget to review this PR now.
Coding agent: Codex
Codex version: codex-cli 0.144.6
Model: gpt-5.6-sol
Reasoning effort: xhigh
| selected_traj = [ | ||
| trajs[np.random.randint(3, len(trajs) - 1)] for _ in range(4) | ||
| ] | ||
| rng = np.random.default_rng() |
There was a problem hiding this comment.
P2 — Preserve reproducible CALYPSO frame selection. This fresh entropy-seeded generator likewise ignores np.random.seed(), while parse_traj() offers no seed or generator parameter. Callers can no longer reproduce which frames are selected. Accept an explicit generator/seed or retain the legacy seeded global call with a targeted Ruff exemption, and add a deterministic selection test.
Codex quota is about to reset, so I am using the remaining token budget to review this PR now.
Coding agent: Codex
Codex version: codex-cli 0.144.6
Model: gpt-5.6-sol
Reasoning effort: xhigh
njzjz-bot
left a comment
njzjz-bot
left a comment
There was a problem hiding this comment.
Reviewed this njzjz-authored draft as requested. The Ruff cleanup has one workflow-breaking dflow identity comparison, one public-export regression, and two reproducibility regressions from unseeded local generators. The branch is also conflicting with current master (14 content conflicts) and has no check runs/statuses on its current head.
Codex quota is about to reset, so I am using the remaining token budget to review this PR now.
Coding agent: Codex
Codex version: codex-cli 0.144.6
Model: gpt-5.6-sol
Reasoning effort: xhigh
2 participants
No description provided.