feat: merge metrics into structured.jsonl, add panther logs metrics CLI#112
Open
ElNiak wants to merge 60 commits into
Open
feat: merge metrics into structured.jsonl, add panther logs metrics CLI#112ElNiak wants to merge 60 commits into
ElNiak wants to merge 60 commits into
Conversation
![sourcery-ai[bot]](https://avatars.githubusercontent.com/in/48477?s=60&v=4){kind=small}
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
Contributor
There was a problem hiding this comment.
Pull request overview
This PR consolidates metrics/event/log outputs into a single structured.jsonl stream and introduces new CLI and reporting utilities to query and summarize those structured records.
Changes:
- Reworks logging to use a JSONL structured file handler and adds context propagation via
LogContext. - Adds post-hoc querying/reporting tooling (
panther logs, RCA, timeline rendering, artifact browsing, output index). - Refactors parts of the event/observer system to use factory methods and broader
BaseEventhandler signatures.
Reviewed changes
Copilot reviewed 76 out of 129 changed files in this pull request and generated 12 comments.
Show a summary per file
| File | Description |
|---|---|
| panther/core/utils/logger_factory.py | Switches file logging to structured JSONL and adds feature/context injection + console-level setter. |
| panther/core/utils/log_context.py | Introduces contextvars-backed context propagation for structured logs/events. |
| panther/core/utils/feature_logger_mixin.py | Removes mixin and keeps a lightweight feature-logger accessor. |
| panther/core/utils/console_formatter.py | Adds a context-aware console formatter and banner helper. |
| panther/core/utils/init.py | Updates utils public API exports for new logging utilities. |
| panther/core/test_cases/test_case_impl.py | Pushes test_id log context during test execution. |
| panther/core/test_cases/mixins/service_management.py | Adds per-service/per-phase log context blocks during service lifecycle. |
| panther/core/reporting/timeline_renderer.py | Adds timeline rendering over structured logs. |
| panther/core/reporting/root_cause_analyzer.py | Adds root-cause analysis over structured logs using failure patterns. |
| panther/core/reporting/log_query_engine.py | Implements streaming JSONL querying with composable filters (incl. metrics). |
| panther/core/reporting/failure_patterns.py | Defines built-in failure pattern catalog for RCA. |
| panther/core/reporting/experiment_reporter.py | Integrates RCA + artifact inventory into generated reports. |
| panther/core/reporting/artifact_browser.py | Adds artifact discovery via output index or directory walk. |
| panther/core/reporting/init.py | Exposes new reporting utilities in the package API. |
| panther/core/outputs/output_index.py | Adds a thread-safe output manifest builder (output_index.json). |
| panther/core/outputs/output_aggregator.py | Registers collected outputs into the output index when provided. |
| panther/core/outputs/init.py | Exports OutputIndexBuilder from outputs package. |
| panther/core/observer/workflow/init.py | Docstring normalization only. |
| panther/core/observer/impl/storage_observer.py | Updates typed handlers to accept BaseEvent and adjusts imports/docs. |
| panther/core/observer/impl/state_observer.py | Simplifies imports and broadens handler signatures to BaseEvent. |
| panther/core/observer/impl/metrics_observer.py | Broadens some typed handlers to BaseEvent and simplifies imports. |
| panther/core/observer/impl/logger_observer.py | Broadens error handlers to BaseEvent and simplifies imports. |
| panther/core/observer/impl/experiment_observer.py | Replaces class-based dispatch with (entity_type, event_name) dispatch. |
| panther/core/observer/impl/event_stream_recorder.py | Adds observer that appends all events into structured.jsonl. |
| panther/core/observer/impl/command_audit_observer.py | Broadens handler signatures and changes supported event type registration. |
| panther/core/observer/impl/init.py | Exports EventStreamRecorder. |
| panther/core/observer/factory.py | Registers event_stream observer type. |
| panther/core/metrics/resource_monitor.py | Docstring normalization only. |
| panther/core/metrics/metric_types.py | Introduces new metric dataclasses. |
| panther/core/metrics/enums.py | Docstring normalization only. |
| panther/core/experiment_observer.py | Registers EventStreamRecorder for experiments. |
| panther/core/events/test/emitter.py | Refactors test emitter to use TestEvent factory methods instead of subclasses. |
| panther/core/events/test/init.py | Reduces exports to the new factory-based API. |
| panther/core/events/step/events.py | Refactors step events to factory methods and removes most subclasses. |
| panther/core/events/step/emitter.py | Updates step emitter to use StepEvent factories. |
| panther/core/events/step/init.py | Reduces exports to the new factory-based API. |
| panther/core/events/service/emitter.py | Refactors service emitter to use ServiceEvent factory methods. |
| panther/core/events/service/init.py | Reduces exports to the new factory-based API. |
| panther/core/events/plugin/emitter.py | Refactors plugin emitter to use PluginEvent factory methods. |
| panther/core/events/plugin/init.py | Reduces exports to the new factory-based API. |
| panther/core/events/experiment/events.py | Refactors experiment events to factory methods; retains a compatibility subclass. |
| panther/core/events/experiment/emitter.py | Updates experiment emitter to use ExperimentEvent factories. |
| panther/core/events/experiment/init.py | Reduces exports to the new factory-based API. |
| panther/core/events/environment/emitter.py | Refactors environment emitter to use EnvironmentEvent factory methods. |
| panther/core/events/environment/init.py | Reduces exports to the new factory-based API. |
| panther/core/events/assertion/events.py | Refactors assertion events to factory methods and removes subclasses. |
| panther/core/events/assertion/emitter.py | Updates assertion emitter to use AssertionEvent factories. |
| panther/core/events/assertion/init.py | Reduces exports to the new factory-based API. |
| panther/core/events/init.py | Shrinks public events API surface to factory-based event classes. |
| panther/core/docker_builder/plugin_mixin/service_manager_docker_mixin.py | Adds user-visible CLI output for cached images. |
| panther/core/docker_builder/platform_detection.py | Adds platform/buildx/build-mode detection and validation helpers. |
| panther/core/command_processor/mixins/modification_mixin.py | Removes command modification mixin. |
| panther/core/command_processor/mixins/init.py | Updates mixin exports after removing modification mixin. |
| panther/core/command_processor/init.py | Updates docs/exports after removing modification mixin. |
| panther/config/core/validators/init.py | Replaces lazy __getattr__ exports with eager imports. |
| panther/config/core/models/global_config.py | Replaces debug_file_logging with structured_log_file config field. |
| panther/cli/core/main.py | Registers new CLI command groups (logs/report/debug/build-metrics). |
| panther/cli/commands/run.py | Adds --debug and passes computed console level into experiment runner. |
| panther/cli/commands/logs.py | Adds panther logs query/errors/metrics subcommands over structured logs. |
| panther/cli/commands/debug.py | Adds developer introspection CLI for events/observers. |
| panther/cli/commands/build_metrics.py | Adds CLI to view/export build/test metrics stored in .panther-metrics. |
| experiment-config/base/experiment_config_example_minimal_docker_no_buildx.yaml | Removes debug_file_logging example field. |
| experiment-config/base/experiment_config_example_minimal_docker.yaml | Removes debug_file_logging example field. |
| experiment-config/base/experiment_config_example_minimal.yaml | Removes debug_file_logging example field. |
You can also share your feedback on Copilot code review. Take the survey.
Comment on lines
+196
to
+198
| # Push log context for this test; will be popped in the finally block | ||
| _ctx_token = log_context(test_id=self.test_name) | ||
| _ctx_token.__enter__() |
Comment on lines
+443
to
+444
| # Pop log context for this test | ||
| _ctx_token.__exit__(None, None, None) |
Comment on lines
+132
to
+133
| record.context_tag = _build_context_tag() # type: ignore[attr-defined] | ||
| record.short_module = _short_module(record.module) # type: ignore[attr-defined] |
Comment on lines
+81
to
+83
| with self._lock: | ||
| with open(self.output_path, "a", encoding="utf-8") as fh: | ||
| fh.write(line + "\n") |
Comment on lines
1
to
5
| """Storage Observer Module.""" | ||
|
|
||
| from typing import Any, Dict, List, Optional | ||
|
|
||
| """ |
| file_level = logging.DEBUG if debug_file_logging else console_level | ||
|
|
||
| # Structured JSONL file handler (replaces old text file handler) | ||
| structured_path = cls._structured_log_file or cls._config.get("output_file") |
| before: Include records with ``ts`` strictly before this datetime. | ||
| correlation_id: Match records with this ``correlation_id``. | ||
| message_pattern: Regex applied to the ``message`` field. | ||
| sources: Filter by ``source`` (``"logging"`` or ``"event"``). |
Comment on lines
+170
to
+184
| error_records = list(self._engine.query(error_filter)) | ||
|
|
||
| # Also collect event-source records with error-related types | ||
| event_filter = LogFilter(sources={"event"}) | ||
| for record in self._engine.query(event_filter): | ||
| message = record.get("message", "") | ||
| event_type = record.get("event_type", "") | ||
| # Include events that look error-related | ||
| if any( | ||
| kw in (message + " " + event_type).lower() | ||
| for kw in ("error", "fail", "crash", "timeout", "killed") | ||
| ): | ||
| # Avoid duplicates | ||
| if record not in error_records: | ||
| error_records.append(record) |
Comment on lines
+544
to
+547
| return (self.experiment_dir / "structured.jsonl").is_file() or any( | ||
| self.experiment_dir.rglob("structured.jsonl") | ||
| ) | ||
|
|
| TestTeardownStartedEvent, | ||
| ) | ||
|
|
||
| __all__ = [ |
|
❌ PR Validation Failed Your changes have validation issues. Please check the following:
You can run the quick validation locally with: # Install dependencies
pip install -e .
pip install pytest black isort flake8
# Run validation
black --check tests/ panther/
isort --check-only tests/ panther/
flake8 tests/ panther/ --max-line-length=88
pytest tests/unit/ -n auto -m "not slow" -p no:panther_metricsPlease fix these issues and push again. The CI will automatically re-run. |
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
10 similar comments
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
…d classes Track 1 - God class splits: - Split DockerBuilder into platform_detection, buildx_operations, image_operations - Split MetricsCollector into metric_types, metric_analysis - Split ExperimentManager into experiment_phases, experiment_cleanup Track 2 - Event/observer simplification: - Replace ~130 event subclasses with factory classmethods on 8 base classes - Simplify ITypedObserver from 881 to ~80 lines with dynamic dispatch - Remove all backward-compat alias functions from events - Clean up events/__init__.py from 404 to ~100 lines - Update all emitters to use factory methods directly Track 3 - Validator cleanup: - Remove dead validator exports from pydantic_factories.py - Simplify validators/__init__.py (remove __getattr__ machinery) Net reduction: ~5,600 lines across 40 files. 870 tests passing.
…, and mixin inlining Wire hidden core modules to CLI (report, build-metrics, debug commands), delete orphaned tools and unused mixins, inline single-use mixins into their consumers, replace empty protocol placeholders with intermediate base classes (ClientServerProtocolBase, PeerToPeerProtocolBase). Net: -1,746 LOC, 13 files deleted, 5 created, 10 new CLI commands.
Add LogContext (contextvars-based) and StructuredJsonFormatter that replaces flat text file logs with structured JSONL. Every log record now carries experiment_id, test_id, service_id, phase context automatically via contextvars — zero changes to call sites needed. - New: log_context.py (LogContext dataclass + context manager) - New: structured_formatter.py (JSONL logging.Formatter) - Modified: logger_factory.py (structured file handler replaces text) - Modified: global_config.py (removed debug_file_logging, added structured_log_file) - Removed: debug_file_logging references from configs and experiment_manager
Wire LogContext into experiment lifecycle boundaries so that every log record and event emitted during a phase/test/service carries the relevant identifiers (experiment_id, test_id, service_id, phase). Context propagation locations: - ExperimentManager.initialize_experiments: experiment_id + phase - ExperimentManager.run_tests: experiment_id + test_execution phase - ExperimentManager.run_tests per-test loop: test_id - ExperimentManager.cleanup: experiment_id + cleanup phase - TestCase.run: test_id (via manual __enter__/__exit__) - ServiceManagementMixin: service_id + phase for setup_testers, setup_implementations, prepare_services, teardown_services New EventStreamRecorder observer: - Appends every event to the same structured.jsonl used by StructuredJsonFormatter, interleaving events and log records chronologically in a single timeline file - Uses "source": "event" and "level": "EVENT" to distinguish from logging records - Reads LogContext for experiment/test/service/phase fields - Thread-safe file writes via threading.Lock - Registered as a global observer in ExperimentObserverMixin with low priority (10) so business observers run first Also: - Register EventStreamRecorder in observer factory as "event_stream" - Add deprecation note to StorageObserver about separate JSONL files now being redundant with structured.jsonl (kept for backward compat) - 12 new unit tests covering schema, context propagation, deduplication, thread safety, error handling
Streaming JSONL log query engine (LogQueryEngine, LogFilter) that reads structured.jsonl files without loading them fully into memory. Supports filtering by level, service, test, phase, time range, correlation ID, message regex, and source. CLI commands: panther logs query <dir> --level/--service/--test/--phase/... panther logs errors <dir> (shorthand for --level ERROR,CRITICAL) Both commands support --json (default JSONL) and --human (table) output. 35 unit tests cover all filter combinations, file discovery, malformed input handling, and the count() aggregation.
Create an incremental, thread-safe output index that tracks every file produced during an experiment and serializes it as output_index.json. - New: panther/core/outputs/output_index.py with OutputIndexBuilder, FileEntry, and detect_type_and_format helper - Integrate into ExperimentManager: init at startup, register known files (config, reports, metrics, structured log), flush at cleanup - Extend OutputAggregator with optional output_index_builder param to register collected environment outputs in the manifest - 35 unit tests covering serialization, auto-detection, dedup, thread safety, and aggregator integration
…mands Add TimelineRenderer for chronological swimlane display of structured log entries, and ArtifactBrowser for listing/filtering experiment output files from output_index.json (with directory walk fallback). Register both as `panther report timeline` and `panther report artifacts` CLI subcommands with filtering by test, service, time range, phase, and artifact type.
…matching Add declarative FailurePattern library (8 built-in patterns derived from ErrorCategory) and RootCauseAnalyzer that reads structured.jsonl via LogQueryEngine to produce ranked root-cause findings with log excerpts and actionable suggestions. Extend ExperimentReporter with optional RCA and artifact inventory sections in both JSON and Markdown reports (gracefully skipped when structured.jsonl is absent). Add `panther report diagnose` CLI subcommand with --json/--human output.
Replace the raw colorlog/logging.Formatter in LoggerFactory with a new ConsoleFormatter that provides: - Short HH:MM:SS timestamps (no date for live runs) - Phase/service context from LogContext: (phase|service_id) prefix - Abbreviated module names (strips panther.core. prefix) - Static banner() method for phase separators via click.echo() Wire phase banners into ExperimentManager at each transition: initialization, plugin loading, test execution, and cleanup.
Add LoggerFactory.set_console_level() classmethod that updates all StreamHandler instances (root + named loggers) to a given level while keeping FileHandlers at DEBUG for structured JSONL output. Wire --verbose (DEBUG) and --debug (TRACE) Click options through the run command into ExperimentManager via a new console_level parameter, giving CLI flags final precedence over YAML config. Includes 8 unit tests covering level updates, FileHandler preservation, TRACE support, and idempotency.
Verify format_test_header() and format_test_result() output formatting, emoji control, timing precision, and padding behavior.
Add format_experiment_summary() that displays a structured banner after all tests complete, showing pass/fail counts with percentage, duration, output directory, failed test details, and next-step CLI commands. Replaces the simple one-line completion message.
DockerOutputParser now adapts to console verbosity: concise one-liner step labels by default, full instruction text when --verbose/--debug is active. Download/extract progress is suppressed in concise mode. Build start, completion (with duration), and cache-skip messages use click.echo() for consistent structured output.
Replace fragile conditional-then-remove command building with clean append-based construction.
…plugin create_plugin() and create_subplugin() do not accept force_overwrite.
discover_plugins() returns Dict[str, PluginMetadata], not Dict[str, List[str]]. Fix params to use direct dict lookup and scan to group by metadata.type for display.
click.Path() returns str but validate() calls .is_dir() and .is_file(). Add Path() conversion matching the existing pattern in check_deps().
… observer comment - Add missing FastFailHandler import (2 tests crashed with NameError) - Remove module-level DockerBuilder.reset_singleton() (unsafe for xdist) - Add step.* prefix tests for MetricsObserver.is_interested() - Fix stale comment about MetricsObserver event scope
…s, metrics filter) 1.1: Fix Jinja2 whitespace stripping that merged success summary lines 1.2: Silence ~38 unhandled event types with noop handlers + configurable logging 1.3: Broaden MetricsObserver.is_interested() to accept test.* and step.* events 1.4: Group service health tables by test_name in both Jinja2 and fallback reports 1.5: Remove duplicate TestCompleted/TestFailed emissions from ExperimentManager 1.6: Add required/optional pattern distinction for output completeness calculation
TestServiceHealthTestName and TestErrorCategoryMatching were defined twice in test_status_collector.py due to the production merge.
…create commands These options were accepted by `panther create plugin` and `panther create subplugin` but silently ignored — never forwarded to create_plugin()/create_subplugin(). Removes them to eliminate user confusion and clean up the public API surface.
- Change [tool:pytest] to [pytest] in tests/cli/pytest.ini (fixes PytestUnknownMarkWarning for @pytest.mark.unit in CLI tests) - Fix validate() and check_deps() type annotations: click.Path() returns str, not Path
- Add pytestmark = pytest.mark.integration at module level - Fix Mock identity comparison (== -> is) for fast_fail_handler - Replace DockerConnection test: builder enters cache-only mode (client=None), not exception - Remove overly specific message/context assertions from docker build failure tests - Add mock_docker_client fixture to test_docker_client_none_raises_exception (singleton isolation) - Fix test_docker_client_none: catches PantherException (validate_build_prerequisites fires first) - Fix test_docker_build_failure_stops_experiment: test handler directly instead of via build_image - Fix test_configuration_based_fast_fail_behavior: remove critical_only (nonexistent), test real behavior - Fix mock_event_manager: configure cleanup_none_observers() to return 0 (not MagicMock) - Fix TestPluginManagerFastFail: use plugin_factory/create_service_manager (service_factory doesn't exist) - Fix test_experiment_initialization_error: assign experiment_config; catch TestCaseInitializationError
…ed API, mock setup) - Replace removed _deploy_services_non_blocking() / deploy_services() tests with test_deploy_services_monitoring() and test_monitoring_not_active_before_deployment_event() - Fix thread-based tests to use polling with deadline instead of fixed sleeps - Mock execute_docker_command (not _is_service_ready) to match _is_service_ready_with_timeout() internals - Fix stop_monitoring() assertion: now sets monitor_thread=None internally - Fix docker_compose_env fixture: env_type must be "network_environment" not "network" - Fix test_test_case_checks_early_termination: use MagicMock for configs, steps.wait attribute - Add pytestmark = pytest.mark.integration
…ed API, mock setup) - Add pytestmark = pytest.mark.integration - Fix thread timing: use stop_monitoring() internal join instead of sleep - Replace removed _deploy_services_non_blocking/deploy_services API calls - Fix BackgroundServiceMonitor mock: mock execute_docker_command not _is_service_ready - Fix TestCase mock setup with proper patches - Clean up unused imports (threading, TestConfig) - Fix BackgroundServiceMonitor import path (moved to own module)
ElNiak
force-pushed
the
feature/metrics-to-structured-jsonl
branch
from
fab3a5a to
31a44e7
Compare
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
… code) - Add threading.Lock to FastFailHandler for thread-safe error tracking - Add warning log + method guard to _requires_emitter and _emit_event - Remove broken conftest fixtures referencing deleted modules - Fix EventStreamRecorder file handle reset to be inside lock - Return False from LoggerObserver on handler exceptions - Pass global_config to metrics/storage observer factory builders - Move logger assignments below imports (isort compliance)
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
- Delegate EventStreamRecorder.__del__ to close() for thread safety - Guard handle_event against non-service events via entity_type check - Normalize datetimes to UTC before stripping tzinfo in log query engine - Correct _build_context_tag docstring separator to match code - Log ImportError in webapp components instead of silently swallowing - Use lazy log formatting in subprocess_executor - Skip panther_ivy tests when submodule not installed (fixes CI)
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
Wrap webapp/components/forms/__init__.py imports in try/except ImportError so test collection succeeds when nicegui is not installed.
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
…B memory usage Coverage was baked into pyproject.toml addopts, meaning every pytest invocation instrumented 321K lines with branch tracing + generated 3 report formats. This caused ~50GB memory usage on casual test runs. Changes: - Remove --cov* from default addopts (coverage now opt-in) - Set branch = false (halves instrumentation overhead) - Enable parallel = true (distribute across SQLite files with xdist) - Expand omit patterns (webapp, tools, banner, install_templates) - Update CLAUDE.md with explicit coverage commands
|
✅ PR Validation Passed Your changes look good! The quick validation checks have passed:
The full CI pipeline will run when this PR is merged or when targeting the main branches. |
Critical fixes: - Fix UnboundLocalError in run_tests() by initializing counters before try - Fix double LoggerFactory.initialize() so root logger gets JSONL handler - Add explicit EventStreamRecorder.close() in cleanup to prevent data loss - Upgrade record_error defensive catches from DEBUG to WARNING - Add concurrent thread-safety tests for FastFailHandler Important fixes: - Replace bare except:pass with circuit-breaker logging in resource_sampler - Stop returning fabricated zero metrics on resource collection error - Return False from _set_state on exception instead of always True - Fix event.event_type -> event.get_type() in docstring example - Rename stale DebugObserver logger to LoggerObserver, add recursion warning - Remove unused enable_colors/structured_output params from _setup_logging - Add experiment_workflow field to FeatureLogLevelsConfig - Count and warn about skipped malformed JSONL lines in StorageObserver - Narrow subprocess retry exception types to prevent wasted retries Tests: - Add FastFailHandler concurrent thread-safety tests - Add MetricsCollector JSONL write circuit-breaker test - Add dot-separated event name dispatch test - Add build-metrics CLI command tests
…tion - Increase stdout/stderr truncation from 500 to 2000 chars in DockerComposeException for better diagnostics - Wrap NiceGUI UI updates in try/except RuntimeError to handle client disconnection gracefully during experiment execution
Reformat multi-line docstrings to single-line where appropriate (ruff D-rule compliance). Add missing module, class, and method docstrings throughout panther/ and tests/. Net reduction of ~370 lines from docstring compaction.
Replace three independent file-write paths (LoggerFactory FileHandler, EventStreamRecorder lock+handle, MetricsCollector lock+open) with a single JsonlWriter that serializes all structured.jsonl output through one threading.Lock and one persistent file handle. This eliminates the critical bug where interleaved writes from concurrent producers could corrupt the JSONL file. - Add panther/core/utils/jsonl_writer.py (JsonlWriter, JsonlLogHandler) - LoggerFactory: use JsonlLogHandler instead of FileHandler, expose get_jsonl_writer() for other components - EventStreamRecorder: remove own lock/handle/close, delegate to writer - MetricsCollector: remove _jsonl_lock and per-write open(), use writer - ExperimentManager: wire shared writer to metrics and event recorder - Update all affected tests (8 new, 3 test files updated)
Fix ImportError handling (C2/C3), experiment_manager context manager (C4), legacy code removal (4a-4f), error handling improvements (5a-5c), bounded dedup set, consistent global_config injection, docstring accuracy fixes, and strict event filtering. Add test coverage for _requires_emitter decorator and ExperimentObserverMixin. Add pytest.mark.unit markers to 3 test files. Fix .add() → OrderedDict in typed_observer and update test_observer_dispatch to remove _HANDLER_OVERRIDES references.
|
❌ PR Validation Failed Your changes have validation issues. Please check the following:
You can run the quick validation locally with: # Install dependencies
pip install -e .
pip install pytest black isort flake8
# Run validation
black --check tests/ panther/
isort --check-only tests/ panther/
flake8 tests/ panther/ --max-line-length=88
pytest tests/unit/ -n auto -m "not slow" -p no:panther_metricsPlease fix these issues and push again. The CI will automatically re-run. |
Replace fragile __class__ mutation with direct mixin inheritance, use call.kwargs for observer_id extraction instead of positional indexing, and strengthen warning count assertion from >= 1 to >= 2.
|
❌ PR Validation Failed Your changes have validation issues. Please check the following:
You can run the quick validation locally with: # Install dependencies
pip install -e .
pip install pytest black isort flake8
# Run validation
black --check tests/ panther/
isort --check-only tests/ panther/
flake8 tests/ panther/ --max-line-length=88
pytest tests/unit/ -n auto -m "not slow" -p no:panther_metricsPlease fix these issues and push again. The CI will automatically re-run. |
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.
Summary
source="metrics") directly tostructured.jsonlinstead of separatemetrics/directory with CSV/JSON filespanther logs metricsCLI: New subcommand with--nameand--typefilters for querying metricsoutputs/metrics/directoryBuilds on PR #110 (structured JSONL logging) and PR #111 (console UX).
Test plan
structured.jsonlcontains metric entriesmetrics/directory no longer createdpanther logs metricsCLI on real output