feat(A1): 变异执行器+定向初筛+属性/裁判+分数台账（IR-0004 AC-1/2 rev6，卡 .github#322/#323） - #104
Merged
Conversation
…绿；卡 .github#322/#323）
|
Warning Review limit reachedNext included review available in 3 minutes. View limit detailsLimit details: You’ve used all 10 included reviews currently available. Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. Review configuration: ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (9)
✨ Finishing Touches 💡 1🛠️ Fix failing CI checks 💡
🧪 Generate unit tests (beta)
Comment |
PR Summary by QodoAdd A1 mutation/property scoring tools with tamper-evident ledger + weekly workflow
AI Description
Diagram
High-Level Assessment
Files changed (9)
|
Code Review by Qodo
1. Mutmut glob passed as path
|
Comment on lines
+459
to
+462
| cmd = [sys.executable, "-m", "mutmut", "run", | ||
| "--paths-to-mutate", module_glob, "--no-progress"] | ||
| proc = subprocess.run(cmd, cwd=str(work), capture_output=True, text=True, | ||
| timeout=overall_timeout) |
There was a problem hiding this comment.
1. Mutmut glob passed as path 🐞 Bug ≡ Correctness
run_mutmut() passes a file-glob like "src/**/*.py" to mutmut via --paths-to-mutate, but that flag is for source directories/packages, not selecting individual files; with the workflow always installing mutmut, engine=auto will pick mutmut and may fail or score the wrong scope.
Agent Prompt
### Issue description
`run_mutmut()` uses `--paths-to-mutate <module_glob>` where `<module_glob>` is a recursive file glob (e.g. `src/**/*.py`). In mutmut, `paths_to_mutate` is intended to point at source directories/packages needed to run tests, not a file-selection glob. With the weekly workflow installing mutmut, `--engine auto` will select the mutmut engine and this mismatch can break the run or produce an unintended mutation scope.
### Issue Context
- The workflow computes `MODULE_GLOB` as a file glob (e.g. `src/**/*.py`).
- The builtin engine correctly treats it as a glob.
- The mutmut engine should either:
1) translate that glob to one or more source root directories, or
2) ignore `module_glob` for mutmut and rely on mutmut’s own configuration/selection mechanism, or
3) force `--engine builtin` in the scheduled workflow.
### Fix Focus Areas
- pipeline/testing/mutation/run_mutation.py[459-463]
- .github/workflows/mutation-weekly.yml[54-71]
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
Comment on lines
+67
to
+69
| records = _read_records(path) | ||
| prev_hash = records[-1]["hash"] if records else GENESIS | ||
| record = { |
There was a problem hiding this comment.
2. Ledger append doesn’t validate 🐞 Bug ☼ Reliability
append_entry() assumes the last record has a valid "hash" and never validates the existing hash chain before extending, so a corrupted/tampered ledger can cause a KeyError crash or be extended without an explicit integrity failure at append time.
Agent Prompt
### Issue description
`append_entry()` reads prior records and unconditionally uses `records[-1]["hash"]` as `prev_hash`. If the last record is missing `hash` (or the file is otherwise corrupt-but-JSON), this raises `KeyError` (uncaught) instead of a controlled `LedgerError`. Also, `append_entry()` does not verify the existing chain before appending, so callers can extend a tampered ledger unless they remember to call `verify()` separately.
### Issue Context
The workflow currently calls `append` and then `verify`, but the CLI/tool should be robust and fail-closed on its own. The module docstring claims it “refuses to extend a structurally corrupt ledger”.
### Fix Focus Areas
- pipeline/testing/mutation/ledger.py[61-82]
- pipeline/testing/mutation/ledger.py[85-125]
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
Comment on lines
+403
to
+407
| mutated = mutate_at(source, site["index"]) | ||
| original = work_file.read_text(encoding="utf-8") | ||
| work_file.write_text(mutated, encoding="utf-8", newline="\n") | ||
| run = run_test_suite(work, runner=runner, timeout=per_mutant_timeout) | ||
| work_file.write_text(original, encoding="utf-8", newline="\n") |
There was a problem hiding this comment.
3. Pycache can skew mutations 🐞 Bug ☼ Reliability
The builtin mutation engine repeatedly overwrites the same file and reruns tests without disabling bytecode caching or clearing __pycache__, which can (on coarse mtime filesystems or same-size rewrites) cause Python to reuse stale .pyc and yield incorrect killed/survived results.
Agent Prompt
### Issue description
The builtin engine mutates a file, runs the suite, then restores the file, but it never clears `__pycache__` and doesn’t run Python with `-B`/`PYTHONDONTWRITEBYTECODE`. In repeated runs, Python may load cached bytecode that no longer matches the current source, which can silently misclassify mutants.
### Issue Context
The judge explicitly clears `__pycache__` between mutant/property runs, indicating the project already recognizes cache staleness as a correctness risk.
### Fix Focus Areas
- pipeline/testing/mutation/run_mutation.py[284-306]
- pipeline/testing/mutation/run_mutation.py[391-419]
- pipeline/testing/property/judge.py[116-119]
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
Comment on lines
+432
to
+435
| try: | ||
| _run() | ||
| except Exception as exc: # noqa: BLE001 - hypothesis re-raises the falsification | ||
| return {"name": spec["name"], "property": spec["property"], |
There was a problem hiding this comment.
4. Hypothesis mislabels config errors 🐞 Bug ≡ Correctness
The Hypothesis engine catches all exceptions as "failed", so manifest/configuration errors (ManifestError) become test failures instead of "error", which misreports operator mistakes and breaks parity with the random engine’s error handling.
Agent Prompt
### Issue description
In `_run_hypothesis()`, `ManifestError` thrown from `_check_property()` is caught under a broad `except Exception` and returned as `status: failed`. In `_run_random()`, `ManifestError` is treated as `status: error`. This makes the same bad manifest look like a property falsification under Hypothesis.
### Issue Context
`_check_property()` can raise `ManifestError` at runtime (e.g., invariant properties missing `gen.params.check`). With Hypothesis installed, `engine=auto` selects Hypothesis and the result classification becomes inconsistent.
### Fix Focus Areas
- pipeline/testing/property/invariants.py[322-335]
- pipeline/testing/property/invariants.py[384-446]
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
变更（pipeline/testing/{mutation,property}）
55 用例全绿（约 16s 离线）+ CLI 端到端冒烟（20 变异体/60% 分/裁判 8 存活→2 accepted 3 trivial rejected）。
依据
IR-0004 rev6（.github#359）；LLM 只出候选、判定全机械（rev6 边界）。
Cards: Cloudbird-Software/.github#322 Cloudbird-Software/.github#323
依据
ADR-0085（PM 优先范式——LLM 只出候选、判定全机械）；IR-0004 rev6（.github#359）。