378 feature request make sure that results obtained from different benchmark versions are not compared#504
Open
dodu94 wants to merge 11 commits intodevelopingfrom
Conversation
…esults-obtained-from-different-benchmark-versions-are-not-compared
…re-that-results-obtained-from-different-benchmark-versions-are-not-compared
Contributor
WalkthroughThe PR lazy-loads GlobalStatus in JadeApp, moves raw-results and simulations parsing to cached properties, adds benchmark-version validation for requested libraries (raising VersionInconsistencyError on mismatch), refactors simulation success checkers into classes, updates run/post-processing to use the new checkers and validation, and adapts tests accordingly. Changes
Sequence Diagram(s)sequenceDiagram
actor User
participant App as JadeApp
participant Status as GlobalStatus
participant Validator as _validate_libs_processing
participant Post as Excel/Atlas Processing
User->>App: post_process(codelibs, benchmark)
App->>Status: access status (lazy-load if _status is None)
Status-->>App: GlobalStatus instance
App->>Status: validate_codelibs(codelibs, benchmark)
Status->>Validator: _validate_libs_processing(code, benchmark, libs)
alt Version mismatch
Validator-->>Status: VersionInconsistencyError
Status-->>App: raise error
App-->>User: post_process fails
else Versions consistent
Validator-->>Status: OK
Status-->>App: validation passed
App->>Post: create outputs & run Excel/Atlas processing
Post-->>App: processing complete
App-->>User: post_process succeeds
end
Estimated code review effort🎯 4 (Complex) | ⏱️ ~50 minutes Possibly related PRs
Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
added 2 commits
March 30, 2026 15:05
Codecov Report❌ Patch coverage is
🚀 New features to boost your workflow:
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 4
🧹 Nitpick comments (1)
tests/config/test_status.py (1)
52-60: Exercise the publicvalidate_codelibs()path in this test.
JadeApp.post_process()calls the public wrapper with a[(CODE, lib), ...]list. Going straight to_validate_libs_processing()bypasses that grouping/dispatch layer, so this can miss bugs in the real production entry point and will break on harmless internal refactors.♻️ Suggested test shape
- with pytest.raises(VersionInconsistencyError): - status._validate_libs_processing( - CODE.MCNP, "Oktavian", ["FENDL 3.2c", "ENDFB-VIII.0"] - ) - status._validate_libs_processing( - CODE.MCNP, "Sphere", ["FENDL 3.2c", "ENDFB-VIII.0"] - ) + with pytest.raises(VersionInconsistencyError): + status.validate_codelibs( + [(CODE.MCNP, "FENDL 3.2c"), (CODE.MCNP, "ENDFB-VIII.0")], + "Oktavian", + ) + status.validate_codelibs( + [(CODE.MCNP, "FENDL 3.2c"), (CODE.MCNP, "ENDFB-VIII.0")], + "Sphere", + )🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@tests/config/test_status.py` around lines 52 - 60, The test should exercise the public entrypoint validate_codelibs instead of calling the private helper _validate_libs_processing directly; replace the two calls to status._validate_libs_processing(...) with calls that pass a list of (CODE, lib) pairs to status.validate_codelibs([(CODE.MCNP, "Oktavian")]) wrapped in pytest.raises(VersionInconsistencyError) for the failing case, and status.validate_codelibs([(CODE.MCNP, "Sphere")]) for the succeeding case so the test follows the same grouping/dispatch path used by JadeApp.post_process().
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@src/jade/app/app.py`:
- Around line 58-69: The cached GlobalStatus returned by JadeApp.status can
become stale after write operations; update JadeApp so that methods that mutate
the simulations/raw trees (e.g., run_benchmarks, raw_process, and any other
writer methods) invalidate the cache by setting self._status = None after
completing their write work, and ensure any helper functions that modify the
tree do the same; locate the status property and add cache-clearing calls in the
implementations of run_benchmarks and raw_process (and post_process if it
mutates data) so subsequent accesses to JadeApp.status construct a fresh
GlobalStatus.
In `@src/jade/config/status.py`:
- Around line 52-55: The raw-data discovery currently requires metadata.json and
lets missing or malformed metadata raise KeyError later; update the logic in the
raw_data getter and in _parse_raw_results_folder to tolerate absent/malformed
metadata by returning None or an empty dict for self._raw_metadata when
metadata.json is missing or invalid, and change _validate_libs_processing to
check for the presence of keys like "benchmark_version" via .get or explicit
presence checks on self._raw_metadata before indexing so it raises a clear
ConsistencyError (or similar) with a descriptive message instead of a raw
KeyError; ensure raw_results_path, _raw_metadata, _raw_data,
_parse_raw_results_folder, and _validate_libs_processing are the symbols updated
so availability queries succeed and validation fails cleanly.
- Around line 46-48: The cached attribute self._simulations (accessed by the
simulations property) is never refreshed after first access, so methods like
was_simulated() and get_successful_simulations() can become stale; modify the
simulations property (and the similar block at lines ~66-72) to add an explicit
refresh path (e.g., an optional force_refresh flag or refresh method) that calls
self._parse_simulations_folder(self.simulations_path) to repopulate
self._simulations when update_raw_results() runs or when callers request a fresh
scan; update GlobalStatus to call that refresh (or pass force_refresh=True)
before was_simulated() / get_successful_simulations() use the cache so the
folder is rescanned for new/removed simulations.
In `@src/jade/helper/aux_functions.py`:
- Around line 72-126: The checkers lost run-directory context because
SimulationChecker.check_success(files: list[str]) only receives filenames;
change the abstract signature to check_success(files: list[str], run_dir: str |
Path) -> bool (or add an optional run_dir: str | Path = None) on
SimulationChecker and update all concrete implementations
(MCNPChecker.check_success, OpenMCChecker.check_success,
SerpentChecker.check_success, D1SChecker.check_success) to accept and use
run_dir when resolving/validating outputs (keeping existing behavior if run_dir
is None for backward compatibility); then update callers (e.g., the code paths
in src/jade/config/status.py and src/jade/run/benchmark.py) to pass the run
directory into check_success and update docstrings/type hints accordingly.
---
Nitpick comments:
In `@tests/config/test_status.py`:
- Around line 52-60: The test should exercise the public entrypoint
validate_codelibs instead of calling the private helper
_validate_libs_processing directly; replace the two calls to
status._validate_libs_processing(...) with calls that pass a list of (CODE, lib)
pairs to status.validate_codelibs([(CODE.MCNP, "Oktavian")]) wrapped in
pytest.raises(VersionInconsistencyError) for the failing case, and
status.validate_codelibs([(CODE.MCNP, "Sphere")]) for the succeeding case so the
test follows the same grouping/dispatch path used by JadeApp.post_process().
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: cf6c71ca-11de-4756-b1b1-8a13ced0ecf1
📒 Files selected for processing (9)
src/jade/app/app.pysrc/jade/config/status.pysrc/jade/helper/aux_functions.pysrc/jade/helper/errors.pysrc/jade/post/sim_output.pysrc/jade/run/benchmark.pytests/app/test_app.pytests/config/test_status.pytests/dummy_structure/raw_data/_mcnp_-_ENDFB-VIII.0_/Oktavian/metadata.json
Comment on lines
+58
to
+69
| self._status = None | ||
|
|
||
| @property | ||
| def status(self) -> GlobalStatus: | ||
| """Lazy-load the global status on first access.""" | ||
| if self._status is None: | ||
| logging.info("Initializing the global status (lazy-loaded)") | ||
| self._status = GlobalStatus( | ||
| simulations_path=self.tree.simulations, | ||
| raw_results_path=self.tree.raw, | ||
| ) | ||
| return self._status |
Contributor
There was a problem hiding this comment.
Invalidate the cached GlobalStatus after write operations.
This property memoizes one GlobalStatus snapshot for the lifetime of JadeApp. After run_benchmarks() or raw_process() changes the simulations/raw trees, later calls still consult stale simulations / raw_data unless callers refresh manually — tests/app/test_app.py Lines 63-72 now has to do that explicitly. A same-session raw_process() → post_process() flow can therefore miss the raw results it just produced.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@src/jade/app/app.py` around lines 58 - 69, The cached GlobalStatus returned
by JadeApp.status can become stale after write operations; update JadeApp so
that methods that mutate the simulations/raw trees (e.g., run_benchmarks,
raw_process, and any other writer methods) invalidate the cache by setting
self._status = None after completing their write work, and ensure any helper
functions that modify the tree do the same; locate the status property and add
cache-clearing calls in the implementations of run_benchmarks and raw_process
(and post_process if it mutates data) so subsequent accesses to JadeApp.status
construct a fresh GlobalStatus.
Comment on lines
+46
to
+48
| if self._simulations is None: | ||
| self._simulations = self._parse_simulations_folder(self.simulations_path) | ||
| return self._simulations |
Contributor
There was a problem hiding this comment.
Add a refresh path for cached simulation state.
After the first simulations access, this instance never rescans simulations_path. update_raw_results() only refreshes raw outputs, so was_simulated() and get_successful_simulations() can stay stale for the lifetime of a long-lived GlobalStatus.
Possible fix
+ def update_simulations(self) -> None:
+ """Update the simulations by re-parsing the simulations folder."""
+ self._simulations = self._parse_simulations_folder(self.simulations_path)
+
def update_raw_results(self) -> None:
"""Update the raw results by re-parsing the raw results folder. It should be used
after processing new raw results to update the status.
"""
self._raw_data, self._raw_metadata = self._parse_raw_results_folder(
self.raw_results_path
)Also applies to: 66-72
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@src/jade/config/status.py` around lines 46 - 48, The cached attribute
self._simulations (accessed by the simulations property) is never refreshed
after first access, so methods like was_simulated() and
get_successful_simulations() can become stale; modify the simulations property
(and the similar block at lines ~66-72) to add an explicit refresh path (e.g.,
an optional force_refresh flag or refresh method) that calls
self._parse_simulations_folder(self.simulations_path) to repopulate
self._simulations when update_raw_results() runs or when callers request a fresh
scan; update GlobalStatus to call that refresh (or pass force_refresh=True)
before was_simulated() / get_successful_simulations() use the cache so the
folder is rescanned for new/removed simulations.
Comment on lines
+52
to
+55
| if self._raw_data is None: | ||
| self._raw_data, self._raw_metadata = self._parse_raw_results_folder( | ||
| self.raw_results_path | ||
| ) |
Contributor
There was a problem hiding this comment.
Make raw-data discovery tolerant to missing metadata, then fail cleanly during validation.
raw_data now depends on metadata.json being present for every non-exp benchmark, so folders without that file will break simple availability queries. _validate_libs_processing() then indexes "benchmark_version" directly, which turns malformed metadata into a raw KeyError instead of a clear consistency error.
Possible fix
- available_raw_data[(CODE(code), lib, benchmark)] = os.listdir(
- bench_path
- )
+ available_raw_data[(CODE(code), lib, benchmark)] = [
+ entry for entry in os.listdir(bench_path)
+ if entry != "metadata.json"
+ ]
if lib != "exp": # for the experiments we don't have metadata
- with open(os.path.join(bench_path, "metadata.json")) as infile:
- metadata[(CODE(code), lib, benchmark)] = json.load(infile)
+ metadata_file = Path(bench_path, "metadata.json")
+ if metadata_file.is_file():
+ with metadata_file.open() as infile:
+ metadata[(CODE(code), lib, benchmark)] = json.load(infile)
...
versions = {}
for lib in libs:
- versions[lib] = self.raw_metadata[(code, lib, benchmark)][
- "benchmark_version"
- ]
+ version = self.raw_metadata.get((code, lib, benchmark), {}).get(
+ "benchmark_version"
+ )
+ if version is None:
+ raise VersionInconsistencyError(
+ f"Missing benchmark_version metadata for {code} / {lib} / {benchmark}"
+ )
+ versions[lib] = versionAlso applies to: 60-63, 144-146, 292-294
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@src/jade/config/status.py` around lines 52 - 55, The raw-data discovery
currently requires metadata.json and lets missing or malformed metadata raise
KeyError later; update the logic in the raw_data getter and in
_parse_raw_results_folder to tolerate absent/malformed metadata by returning
None or an empty dict for self._raw_metadata when metadata.json is missing or
invalid, and change _validate_libs_processing to check for the presence of keys
like "benchmark_version" via .get or explicit presence checks on
self._raw_metadata before indexing so it raises a clear ConsistencyError (or
similar) with a descriptive message instead of a raw KeyError; ensure
raw_results_path, _raw_metadata, _raw_data, _parse_raw_results_folder, and
_validate_libs_processing are the symbols updated so availability queries
succeed and validation fails cleanly.
Comment on lines
+72
to
+126
| class SimulationChecker(ABC): | ||
| """Abstract base class for simulation success checkers. | ||
|
|
||
| All code-specific checkers must inherit from this class and implement | ||
| the check_success method with the same signature. | ||
| """ | ||
|
|
||
| def check_run_openmc(folder: PathLike) -> bool: | ||
| """check if openmc run was successful""" | ||
| try: | ||
| OpenMCSimOutput.retrieve_file(folder) | ||
| return True | ||
| except FileNotFoundError: | ||
| return False | ||
| @abstractmethod | ||
| def check_success(self, files: list[str]) -> bool: | ||
| """Check if a simulation run was successful. | ||
|
|
||
| Parameters | ||
| ---------- | ||
| files : list[str] | ||
| List of output files to check for success. | ||
|
|
||
| Returns | ||
| ------- | ||
| bool | ||
| True if the simulation completed successfully, False otherwise. | ||
| """ | ||
| pass | ||
|
|
||
|
|
||
| class MCNPChecker(SimulationChecker): | ||
| """Checker for MCNP simulations.""" | ||
|
|
||
| def check_success(self, files: list[str]) -> bool: | ||
| """Check if MCNP run was successful by verifying output files exist.""" | ||
| return MCNPSimOutput.is_successfully_simulated(files) | ||
|
|
||
|
|
||
| class OpenMCChecker(SimulationChecker): | ||
| """Checker for OpenMC simulations.""" | ||
|
|
||
| def check_success(self, files: list[str]) -> bool: | ||
| """Check if OpenMC run was successful by verifying output files exist.""" | ||
| return OpenMCSimOutput.is_successfully_simulated(files) | ||
|
|
||
|
|
||
| class SerpentChecker(SimulationChecker): | ||
| """Checker for Serpent simulations.""" | ||
|
|
||
| def check_success(self, files: list[str]) -> bool: | ||
| """Check if Serpent run was successful.""" | ||
| # TODO implement the logic to check if the Serpent run was successful | ||
| raise NotImplementedError("Serpent checker not yet implemented") | ||
|
|
||
| def check_run_serpent(folder: PathLike) -> bool: | ||
| # TODO implement the logic to check if the Serpent run was successful | ||
| raise NotImplementedError | ||
|
|
||
| class D1SChecker(SimulationChecker): | ||
| """Checker for D1S simulations.""" | ||
|
|
||
| def check_run_d1s(folder: PathLike) -> bool: | ||
| """check if d1s run was successful""" | ||
| return check_run_mcnp(folder) | ||
| def check_success(self, files: list[str]) -> bool: | ||
| """Check if D1S run was successful (uses same logic as MCNP).""" | ||
| return MCNPChecker().check_success(files) |
Contributor
There was a problem hiding this comment.
Keep the run directory in check_success().
The new files: list[str] contract strips the context needed to validate actual completion. As used by src/jade/config/status.py Lines 94-98 and src/jade/run/benchmark.py Lines 441-445, these checkers now have to infer success from filename presence alone, which can mark interrupted runs with leftover artifacts as complete and then skip reruns or post-process bad outputs.
💡 Suggested interface direction
class SimulationChecker(ABC):
"""Abstract base class for simulation success checkers.
@@
`@abstractmethod`
- def check_success(self, files: list[str]) -> bool:
+ def check_success(
+ self, run_dir: PathLike, files: list[str] | None = None
+ ) -> bool:
"""Check if a simulation run was successful.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@src/jade/helper/aux_functions.py` around lines 72 - 126, The checkers lost
run-directory context because SimulationChecker.check_success(files: list[str])
only receives filenames; change the abstract signature to check_success(files:
list[str], run_dir: str | Path) -> bool (or add an optional run_dir: str | Path
= None) on SimulationChecker and update all concrete implementations
(MCNPChecker.check_success, OpenMCChecker.check_success,
SerpentChecker.check_success, D1SChecker.check_success) to accept and use
run_dir when resolving/validating outputs (keeping existing behavior if run_dir
is None for backward compatibility); then update callers (e.g., the code paths
in src/jade/config/status.py and src/jade/run/benchmark.py) to pass the run
directory into check_success and update docstrings/type hints accordingly.
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (1)
src/jade/config/status.py (1)
314-321: RedundantCODEconversion.Line 316 converts
codetoCODE(code), but the type hint on line 302 declarescodelibs: list[tuple[CODE, str]], indicatingcodeshould already be aCODEenum. If callers may pass strings, consider updating the type hint; otherwise, remove the redundant conversion.♻️ Proposed simplification (if type hint is accurate)
for code, lib in codelibs: - code = CODE(code) if lib == "exp": # for the experiments we don't have metadata continue if code not in libs:🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/jade/config/status.py` around lines 314 - 321, The loop is redundantly converting code to CODE(code) even though codelibs is typed as list[tuple[CODE, str]]; remove the unnecessary conversion in the for loop (i.e., stop reassigning code = CODE(code)) so the existing CODE instance from codelibs is used directly, or if callers may pass strings instead of CODE update the codelibs type hint and add validation/conversion at the call sites; target the for loop that iterates over codelibs and the use of CODE to decide which path to take.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@src/jade/config/status.py`:
- Around line 301-312: The validate_codelibs method's parameter is named
benchmarks but its type hint and docstring describe a single benchmark; rename
the parameter to benchmark to match the docstring and type hint and update any
internal uses and callers accordingly (search for validate_codelibs and change
the parameter name in its signature and all invocations to benchmark) so
documentation and code are consistent.
- Around line 285-288: The docstring is incorrect: update the docstring for the
function that raises VersionInconsistencyError (the function documented in this
block) to state that it returns None on success and to include a Raises section
documenting VersionInconsistencyError instead of claiming it returns bool;
specifically change the Returns section to "None" (or remove it) and add a
"Raises\n-------\nVersionInconsistencyError: when post-processing cannot be
performed" so the docstring matches the actual behavior.
---
Nitpick comments:
In `@src/jade/config/status.py`:
- Around line 314-321: The loop is redundantly converting code to CODE(code)
even though codelibs is typed as list[tuple[CODE, str]]; remove the unnecessary
conversion in the for loop (i.e., stop reassigning code = CODE(code)) so the
existing CODE instance from codelibs is used directly, or if callers may pass
strings instead of CODE update the codelibs type hint and add
validation/conversion at the call sites; target the for loop that iterates over
codelibs and the use of CODE to decide which path to take.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 61633b82-8572-41db-b231-8cba6472f1fe
📒 Files selected for processing (1)
src/jade/config/status.py
Comment on lines
+285
to
+288
| Returns | ||
| ------- | ||
| bool | ||
| True if the post-processing can be performed, False otherwise. |
Contributor
There was a problem hiding this comment.
Docstring incorrectly states return type.
The docstring states "Returns -> bool: True if the post-processing can be performed" but the method returns None or raises VersionInconsistencyError. Update the docstring to reflect the actual behavior.
📝 Proposed docstring fix
- Returns
- -------
- bool
- True if the post-processing can be performed, False otherwise.
+ Raises
+ ------
+ VersionInconsistencyError
+ If the benchmark versions are not consistent across the requested libraries.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@src/jade/config/status.py` around lines 285 - 288, The docstring is
incorrect: update the docstring for the function that raises
VersionInconsistencyError (the function documented in this block) to state that
it returns None on success and to include a Raises section documenting
VersionInconsistencyError instead of claiming it returns bool; specifically
change the Returns section to "None" (or remove it) and add a
"Raises\n-------\nVersionInconsistencyError: when post-processing cannot be
performed" so the docstring matches the actual behavior.
Comment on lines
+301
to
+312
| def validate_codelibs( | ||
| self, codelibs: list[tuple[CODE, str]], benchmarks: str | ||
| ) -> None: | ||
| """Check that a list of codelibs can be post-processed together for a given benchmark. | ||
| This is true if the benchmark version for the requested libs are the same. | ||
|
|
||
| Parameters | ||
| ---------- | ||
| codelibs : list[tuple[CODE, str]] | ||
| list of codelibs to check. Each codelib is a tuple with the code and library. | ||
| benchmark : str | ||
| benchmark name. |
Contributor
There was a problem hiding this comment.
Parameter name and docstring inconsistency.
The parameter is named benchmarks (line 302) but the docstring references benchmark (line 311-312). Since the type hint indicates str (singular benchmark), consider renaming the parameter to benchmark for consistency.
📝 Proposed fix
def validate_codelibs(
- self, codelibs: list[tuple[CODE, str]], benchmarks: str
+ self, codelibs: list[tuple[CODE, str]], benchmark: str
) -> None:
"""Check that a list of codelibs can be post-processed together for a given benchmark.
...
for code, lib_list in libs.items():
- self._validate_libs_processing(code, benchmarks, lib_list)
+ self._validate_libs_processing(code, benchmark, lib_list)🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@src/jade/config/status.py` around lines 301 - 312, The validate_codelibs
method's parameter is named benchmarks but its type hint and docstring describe
a single benchmark; rename the parameter to benchmark to match the docstring and
type hint and update any internal uses and callers accordingly (search for
validate_codelibs and change the parameter name in its signature and all
invocations to benchmark) so documentation and code are consistent.
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
1 participant
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.
Fix #378
At last metadata are used to ensure that during post-processing of a specific benchmark (and code) the libraries results requested were all obtained from the same benchmark version.
To be reviewed after #500.
Summary by CodeRabbit
New Features
Refactor