refactor(compute): remove global mutable state from compute_executor [OMN-738]

.github/workflows/test.yml

@@ -23,18 +23,17 @@ jobs:
   # JOB ORDERING RATIONALE (Fail-Fast Optimization)
   # =============================================================================
   #
-  # Phase 1 - Quick Validation (parallel, ~5-10 min):
-  #   - smoke-test: Basic import/syntax checks
+  # Phase 1 - Quick Validation (parallel, ~2-3 min):
   #   - lint: Code quality (ruff format, ruff check, mypy strict)
   #   - pyright: Type checking (complementary to mypy)
   #   - exports-validation: __all__ exports validation
   #   - docs-validation: Documentation link validation
   #   - node-purity-check: Declarative node purity guarantees
   #   - mypy-validation-scripts: Type check validation scripts
   #
-  # Phase 2 - Expensive Tests (after Phase 1 completes):
+  # Phase 2 - Full Test Suite (after Phase 1 completes):
   #   - test-parallel: 20 splits of full test suite (~3 min each)
-  #     Depends on: smoke-test, lint, pyright, exports-validation
+  #     Depends on: lint, pyright, exports-validation
   #     This ensures type errors and lint failures are caught BEFORE
   #     spawning 20 parallel test runners (saves ~60 runner-minutes on failure)
   #
@@ -47,65 +46,6 @@ jobs:
   #   3. All Phase 1 jobs run in parallel for speed
   # =============================================================================
 
-  # Quick smoke test - fails fast on basic issues (5-10s)
-  smoke-test:
-    name: Smoke Tests
-    runs-on: ubuntu-latest
-    timeout-minutes: 5
-
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-
-      - name: Set up Python
-        uses: actions/setup-python@v5
-        with:
-          python-version: ${{ env.PYTHON_VERSION }}
-          cache: 'pip'
-
-      - name: Install Poetry
-        uses: snok/install-poetry@v1
-        with:
-          version: ${{ env.POETRY_VERSION }}
-          virtualenvs-create: true
-          virtualenvs-in-project: true
-
-      # Note: omnibase_spi is now public - no auth required
-
-      - name: Load cached venv
-        id: cached-poetry-dependencies
-        uses: actions/cache@v4
-        with:
-          path: .venv
-          key: >-
-            venv-${{ runner.os }}-${{ env.PYTHON_VERSION }}-
-            ${{ hashFiles('**/poetry.lock') }}-
-            ${{ env.CACHE_VERSION }}
-          restore-keys: |
-            venv-${{ runner.os }}-${{ env.PYTHON_VERSION }}-${{ hashFiles('**/poetry.lock') }}-${{ env.CACHE_VERSION }}
-
-      - name: Install dependencies
-        if: steps.cached-poetry-dependencies.outputs.cache-hit != 'true'
-        run: |
-          poetry install --no-interaction --no-root || {
-            echo "❌ Dependency install failed"
-            exit 1
-          }
-
-      - name: Install project
-        run: |
-          poetry install --no-interaction || {
-            echo "❌ Project install failed"
-            exit 1
-          }
-
-      - name: Run smoke tests
-        run: |
-          poetry run pytest tests/unit/enums tests/unit/errors \
-            --maxfail=5 \
-            -x \
-            --tb=short
-
   # Parallel test execution - split into 20 groups for speed and memory management
   # Split strategy rationale:
   #   - 12,198 total tests / 20 splits = ~610 tests/split
@@ -115,15 +55,14 @@ jobs:
   #   - Shorter per-split runtime reduces probability of external runner shutdown
   #
   # Dependency rationale (fail-fast optimization):
-  #   - smoke-test: Catches basic import/syntax errors before expensive splits
   #   - lint: Catches formatting, linting, and mypy errors early
   #   - pyright: Catches type errors that mypy might miss
   #   - exports-validation: Ensures __all__ exports are valid before tests
-  #   These 4 jobs run in parallel (~10 min total) and must all pass before
+  #   These 3 jobs run in parallel (~3 min total) and must all pass before
   #   spawning 20 test runners. This saves ~60 runner-minutes on validation failures.
   test-parallel:
     name: Tests (Split ${{ matrix.split }}/20)
-    needs: [smoke-test, lint, pyright, exports-validation]
+    needs: [lint, pyright, exports-validation]
     runs-on: ubuntu-latest
     timeout-minutes: 30
     strategy:
@@ -205,7 +144,7 @@ jobs:
           path: junit-${{ matrix.split }}.xml
           retention-days: 7
 
-  # Phase 1 validation job - runs in parallel with smoke-test, pyright, exports-validation
+  # Phase 1 validation job - runs in parallel with pyright, exports-validation
   # Must complete before test-parallel starts (fail-fast optimization)
   lint:
     name: Code Quality
@@ -259,7 +198,7 @@ jobs:
       - name: Run mypy type checking (strict)
         run: poetry run mypy src/omnibase_core
 
-  # Phase 1 validation job - runs in parallel with smoke-test, lint, exports-validation
+  # Phase 1 validation job - runs in parallel with lint, exports-validation
   # Must complete before test-parallel starts (fail-fast optimization)
   # Pyright complements mypy (in lint job) - catches different type error categories
   pyright:
@@ -306,7 +245,7 @@ jobs:
       - name: Run pyright type checking
         run: poetry run pyright src/omnibase_core
 
-  # Phase 1 validation job - runs in parallel with smoke-test, lint, pyright
+  # Phase 1 validation job - runs in parallel with lint, pyright
   # Must complete before test-parallel starts (fail-fast optimization)
   # Quick check (~5 min) ensures __all__ exports are valid before expensive test splits
   exports-validation:
@@ -607,13 +546,12 @@ jobs:
   test-summary:
     name: Test Summary
     runs-on: ubuntu-latest
-    needs: [smoke-test, test-parallel, lint, pyright, exports-validation, mypy-validation-scripts, docs-validation, node-purity-check]
+    needs: [test-parallel, lint, pyright, exports-validation, mypy-validation-scripts, docs-validation, node-purity-check]
     if: always()
 
     steps:
       - name: Check test results
         run: |
-          smoke="${{ needs.smoke-test.result }}"
           parallel="${{ needs.test-parallel.result }}"
           lint="${{ needs.lint.result }}"
           pyright="${{ needs.pyright.result }}"
@@ -622,16 +560,14 @@ jobs:
           docs="${{ needs.docs-validation.result }}"
           purity="${{ needs.node-purity-check.result }}"
 
-          if [[ "$smoke" == "success" ]] && \
-             [[ "$parallel" == "success" ]] && \
+          if [[ "$parallel" == "success" ]] && \
              [[ "$lint" == "success" ]] && \
              [[ "$pyright" == "success" ]] && \
              [[ "$exports" == "success" ]] && \
              [[ "$mypy_scripts" == "success" ]] && \
              [[ "$docs" == "success" ]] && \
              [[ "$purity" == "success" ]]; then
             echo "✅ All tests passed successfully"
-            echo "Smoke Tests: ✅ Passed"
             echo "Parallel Tests (20 splits): ✅ Passed"
             echo "Code Quality: ✅ Passed"
             echo "Pyright Type Checking: ✅ Passed"
@@ -642,7 +578,6 @@ jobs:
             exit 0
           else
             echo "❌ Tests failed"
-            echo "Smoke Tests: $smoke"
             echo "Parallel Tests: $parallel"
             echo "Lint: $lint"
             echo "Pyright: $pyright"

docs/architecture/CONTRACT_DRIVEN_NODECOMPUTE_V1_0.md

@@ -52,9 +52,7 @@ The core pipeline executor and transformation functions are designed to be **pur
 - Each `execute_compute_pipeline()` invocation operates on its own data without shared state
 - Safe for concurrent use from multiple threads
 
-**Known Limitation (v1.0)**: The validation step warning deduplication uses module-level state (`_validation_warning_emitted`) to prevent log spam. This is a minor exception to the pure function principle, affecting only warning output (not computation results). The pipeline execution itself remains deterministic and thread-safe.
-
-**TODO(v1.1)**: Refactor warning deduplication to use a context-based approach (e.g., per-execution context) rather than global state. See `compute_executor.py` for details.
+**Note (v1.0)**: Validation steps emit a UserWarning indicating pass-through behavior. Warning deduplication is handled by Python's built-in `warnings` module, maintaining the pure function principle throughout the module.
 
 ---
 

src/omnibase_core/utils/compute_executor.py

@@ -54,13 +54,6 @@
 
 logger = logging.getLogger(__name__)
 
-# Track if validation warning has been emitted for session deduplication
-# TODO(): Refactor to context-based warning tracking instead of global state.
-# This global mutable state is a known exception to the "pure and stateless" principle.
-# It affects only warning output (not computation results) and pipeline execution
-# remains deterministic. See: docs/architecture/CONTRACT_DRIVEN_NODECOMPUTE_V1_0.md
-_validation_warning_emitted = False
-
 from omnibase_core.enums.enum_compute_step_type import EnumComputeStepType
 from omnibase_core.enums.enum_core_error_code import EnumCoreErrorCode
 from omnibase_core.models.compute.model_compute_execution_context import (
@@ -355,16 +348,13 @@ def execute_validation_step(
     # - Add validation error messages with path information
     # See: docs/architecture/NODECOMPUTE_VERSIONING_ROADMAP.md
 
-    # Emit UserWarning once per session for validation steps
-    global _validation_warning_emitted
-    if not _validation_warning_emitted:
-        warnings.warn(
-            "Validation steps are pass-through in v1.0. "
-            "Schema validation will be implemented in v1.1.",
-            UserWarning,
-            stacklevel=2,
-        )
-        _validation_warning_emitted = True
+    # Emit UserWarning for validation steps (Python warnings module deduplicates automatically)
+    warnings.warn(
+        "Validation steps are pass-through in v1.0. "
+        "Schema validation will be implemented in v1.1.",
+        UserWarning,
+        stacklevel=2,
+    )
 
     # Log debug-level message for each step (for troubleshooting)
     logger.debug(

tests/unit/utils/test_compute_executor.py

@@ -453,3 +453,33 @@ def test_logger_exception_includes_exception_type(self) -> None:
             # The logger.exception format string includes type(e).__name__
             log_args = call_args[0][1:]
             assert "ZeroDivisionError" in log_args
+
+
+@pytest.mark.unit
+@pytest.mark.timeout(5)
+class TestValidationStepWarning:
+    """Tests for validation step warning behavior."""
+
+    def test_validation_step_emits_warning(self) -> None:
+        """Test that validation steps emit a UserWarning about v1.0 pass-through.
+
+        Verifies that execute_validation_step emits a warning indicating
+        validation is pass-through in v1.0, informing users that schema
+        validation will be implemented in v1.1.
+        """
+        from omnibase_core.models.transformations.model_validation_step_config import (
+            ModelValidationStepConfig,
+        )
+        from omnibase_core.utils.compute_executor import execute_validation_step
+
+        step = ModelComputePipelineStep(
+            step_name="test_validation",
+            step_type=EnumComputeStepType.VALIDATION,
+            validation_config=ModelValidationStepConfig(schema_ref="test_schema"),
+        )
+
+        with pytest.warns(UserWarning, match="pass-through.*v1.1|v1.1.*pass-through"):
+            result = execute_validation_step(step, {"test": "data"})
+
+        # Verify pass-through behavior returns input unchanged
+        assert result == {"test": "data"}