feat: improve mypyc configurations (#388)

Continue optimizations to Mypyc compilation layer

Author: cofin

.pre-commit-config.yaml

@@ -17,7 +17,7 @@ repos:
       - id: mixed-line-ending
       - id: trailing-whitespace
   - repo: https://github.com/charliermarsh/ruff-pre-commit
-    rev: "v0.15.5"
+    rev: "v0.15.6"
     hooks:
       - id: ruff
         args: ["--fix"]

Makefile

@@ -365,7 +365,7 @@ pgo-local:                                          ## Run full three-stage PGO
 	@echo "${OK} Instrumented wheel built"
 	@echo "${INFO} Stage 2: Running training workload..."
 	@uv pip install dist/*.whl --force-reinstall --no-deps >/dev/null 2>&1
-	@.venv/bin/python -m sqlspec._pgo_training
+	@.venv/bin/python tools/scripts/pgo_training.py
 	@echo "${OK} Training complete"
 	@rm -rf dist/ $(PGO_BUILD_DIR)/build $(PGO_BUILD_DIR)/tmp
 	@echo "${INFO} Stage 3: Building PGO-optimized wheel..."

performance_surface_inventory.json

@@ -0,0 +1,285 @@
+{
+  "task_id": "sqlspec-iab.6.1",
+  "captured_on": "2026-03-14",
+  "summary": {
+    "adapter_count": 16,
+    "bench_harness_driver_count": 4,
+    "regression_gate_driver_count": 1
+  },
+  "benchmark_scripts": {
+    "bench.py": {
+      "drivers": [
+        "aiosqlite",
+        "asyncpg",
+        "duckdb",
+        "sqlite"
+      ],
+      "core_scenarios": [
+        "initialization",
+        "write_heavy",
+        "read_heavy",
+        "iterative_inserts",
+        "repeated_queries"
+      ],
+      "extended_scenarios": [
+        "dict_key_transform",
+        "schema_mapping",
+        "complex_parameters",
+        "thin_path_stress"
+      ]
+    },
+    "bench_gate.py": {
+      "drivers": [
+        "sqlite"
+      ],
+      "gate_scenarios": [
+        "iterative_inserts",
+        "repeated_queries",
+        "write_heavy",
+        "read_heavy"
+      ]
+    },
+    "bench_subsystems.py": {
+      "drivers": [
+        "sqlite"
+      ],
+      "focus_areas": [
+        "SQL object construction",
+        "SQL.compile()",
+        "query-cache prepare/lookup",
+        "parameter processing",
+        "cursor lifecycle",
+        "storage/runtime helpers"
+      ]
+    }
+  },
+  "build_workflows": {
+    "publish.yml": {
+      "tier": "release",
+      "builds_mypyc_wheels": true,
+      "exercises_three_stage_pgo": true,
+      "uses_packaged_training_module": true,
+      "notes": "Release wheels still run `python -m sqlspec._pgo_training` during the three-stage PGO flow."
+    },
+    "test-build.yml": {
+      "tier": "pull_request/workflow_dispatch",
+      "builds_mypyc_wheels": true,
+      "exercises_three_stage_pgo": false,
+      "uses_packaged_training_module": false,
+      "notes": "PR build validation compiles mypyc wheels but does not run reduced three-stage PGO parity."
+    }
+  },
+  "external_infra": {
+    "dma_accelerator": {
+      "reusable_lifecycle_modules": [
+        "postgres",
+        "mysql",
+        "oracle",
+        "sqlserver"
+      ],
+      "root_cli_registered_families": [
+        "postgres"
+      ],
+      "notes": "Reusable lifecycle code exists for PostgreSQL, MySQL, Oracle, and SQL Server, but the root CLI currently mounts only PostgreSQL. MySQL and Oracle still need SQLSpec-facing exposure work before they can be treated as directly reusable perf infrastructure."
+    }
+  },
+  "adapters": [
+    {
+      "name": "adbc",
+      "execution_surfaces": [
+        "sync"
+      ],
+      "config_surface": "sqlspec/adapters/adbc/config.py",
+      "integration_root": "tests/integration/adapters/adbc",
+      "current_infra_family": "bridge/underlying-engine",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "Integration coverage spans PostgreSQL, SQLite, and DuckDB backends, but no dedicated perf-harness workload exists yet."
+    },
+    {
+      "name": "aiosqlite",
+      "execution_surfaces": [
+        "async"
+      ],
+      "config_surface": "sqlspec/adapters/aiosqlite/config.py",
+      "integration_root": "tests/integration/adapters/aiosqlite",
+      "current_infra_family": "file-local",
+      "bench_harness_driver": true,
+      "regression_gate_driver": false,
+      "notes": "Covered in bench.py, but not in bench_gate.py or bench_subsystems.py."
+    },
+    {
+      "name": "asyncmy",
+      "execution_surfaces": [
+        "async"
+      ],
+      "config_surface": "sqlspec/adapters/asyncmy/config.py",
+      "integration_root": "tests/integration/adapters/asyncmy",
+      "current_infra_family": "server-backed",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "MySQL-family integration coverage exists, but there is no current perf-harness workload."
+    },
+    {
+      "name": "asyncpg",
+      "execution_surfaces": [
+        "async"
+      ],
+      "config_surface": "sqlspec/adapters/asyncpg/config.py",
+      "integration_root": "tests/integration/adapters/asyncpg",
+      "current_infra_family": "server-backed",
+      "bench_harness_driver": true,
+      "regression_gate_driver": false,
+      "notes": "Bench coverage exists in bench.py, but the regression gate remains SQLite-only."
+    },
+    {
+      "name": "bigquery",
+      "execution_surfaces": [
+        "sync"
+      ],
+      "config_surface": "sqlspec/adapters/bigquery/config.py",
+      "integration_root": "tests/integration/adapters/bigquery",
+      "current_infra_family": "cloud-managed",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "Integration coverage exists, but there is no current perf or build-path matrix entry for the adapter."
+    },
+    {
+      "name": "cockroach_asyncpg",
+      "execution_surfaces": [
+        "async"
+      ],
+      "config_surface": "sqlspec/adapters/cockroach_asyncpg/config.py",
+      "integration_root": "tests/integration/adapters/cockroach_asyncpg",
+      "current_infra_family": "server-backed",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "Cockroach async coverage exists in integration tests only."
+    },
+    {
+      "name": "cockroach_psycopg",
+      "execution_surfaces": [
+        "sync",
+        "async"
+      ],
+      "config_surface": "sqlspec/adapters/cockroach_psycopg/config.py",
+      "integration_root": "tests/integration/adapters/cockroach_psycopg",
+      "current_infra_family": "server-backed",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "Cockroach psycopg integration coverage exists, but no perf harness tracks either sync or async surfaces."
+    },
+    {
+      "name": "duckdb",
+      "execution_surfaces": [
+        "sync"
+      ],
+      "config_surface": "sqlspec/adapters/duckdb/config.py",
+      "integration_root": "tests/integration/adapters/duckdb",
+      "current_infra_family": "file-local",
+      "bench_harness_driver": true,
+      "regression_gate_driver": false,
+      "notes": "Bench coverage exists in bench.py, but no driver-specific gate or subsystem micro-benchmark exists."
+    },
+    {
+      "name": "mock",
+      "execution_surfaces": [
+        "sync",
+        "async"
+      ],
+      "config_surface": "sqlspec/adapters/mock/config.py",
+      "integration_root": null,
+      "current_infra_family": "mock-only",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "Useful for isolated behavior tests, but not a real performance validation surface."
+    },
+    {
+      "name": "mysqlconnector",
+      "execution_surfaces": [
+        "sync",
+        "async"
+      ],
+      "config_surface": "sqlspec/adapters/mysqlconnector/config.py",
+      "integration_root": "tests/integration/adapters/mysqlconnector",
+      "current_infra_family": "server-backed",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "Integration coverage exists for sync and async paths, but the perf harness does not exercise them yet."
+    },
+    {
+      "name": "oracledb",
+      "execution_surfaces": [
+        "sync",
+        "async"
+      ],
+      "config_surface": "sqlspec/adapters/oracledb/config.py",
+      "integration_root": "tests/integration/adapters/oracledb",
+      "current_infra_family": "server-backed",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "Integration coverage exists for sync and async Oracle paths, but there is no Oracle perf workload in the harness."
+    },
+    {
+      "name": "psqlpy",
+      "execution_surfaces": [
+        "async"
+      ],
+      "config_surface": "sqlspec/adapters/psqlpy/config.py",
+      "integration_root": "tests/integration/adapters/psqlpy",
+      "current_infra_family": "server-backed",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "Integration coverage exists, but the current perf harness does not represent the adapter."
+    },
+    {
+      "name": "psycopg",
+      "execution_surfaces": [
+        "sync",
+        "async"
+      ],
+      "config_surface": "sqlspec/adapters/psycopg/config.py",
+      "integration_root": "tests/integration/adapters/psycopg",
+      "current_infra_family": "server-backed",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "Integration coverage exists for sync and async psycopg paths, but the perf harness still skips them."
+    },
+    {
+      "name": "pymysql",
+      "execution_surfaces": [
+        "sync"
+      ],
+      "config_surface": "sqlspec/adapters/pymysql/config.py",
+      "integration_root": "tests/integration/adapters/pymysql",
+      "current_infra_family": "server-backed",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "Integration coverage exists, but no current perf workload or gate covers the adapter."
+    },
+    {
+      "name": "spanner",
+      "execution_surfaces": [
+        "sync"
+      ],
+      "config_surface": "sqlspec/adapters/spanner/config.py",
+      "integration_root": "tests/integration/adapters/spanner",
+      "current_infra_family": "cloud-managed",
+      "bench_harness_driver": false,
+      "regression_gate_driver": false,
+      "notes": "Integration coverage exists, but cloud-managed validation is not represented in the current perf harness."
+    },
+    {
+      "name": "sqlite",
+      "execution_surfaces": [
+        "sync"
+      ],
+      "config_surface": "sqlspec/adapters/sqlite/config.py",
+      "integration_root": "tests/integration/adapters/sqlite",
+      "current_infra_family": "file-local",
+      "bench_harness_driver": true,
+      "regression_gate_driver": true,
+      "notes": "SQLite is the only adapter covered by bench.py, bench_gate.py, and bench_subsystems.py."
+    }
+  ]
+}

pyproject.toml

@@ -199,8 +199,7 @@ exclude = [
   "sqlspec/adapters/**/data_dictionary.py",   # Cross-module inheritance causes mypyc segfaults
   "sqlspec/observability/_formatting.py",     # Inherits from non-compiled logging.Formatter
   "sqlspec/utils/arrow_helpers.py",           # Arrow operations cause segfaults when compiled
-  "sqlspec/utils/profiling.py",               # Uses sys.setprofile (dynamic, not mypyc compatible)
-  "sqlspec/_pgo_training.py",                 # PGO training script — runs against compiled modules
+  "sqlspec/storage/backends/_iterators.py",  # Async __anext__ + asyncio.to_thread causes mypyc segfault
 ]
 include = [
   "sqlspec/core/**/*.py",                  # Core module
@@ -209,6 +208,9 @@ include = [
   "sqlspec/loader.py",                     # Loader module
   "sqlspec/observability/**/*.py",         # Observability utilities
   "sqlspec/driver/**/*.py",                # Driver module
+  "sqlspec/storage/registry.py",           # Safe storage registry/runtime routing
+  "sqlspec/storage/errors.py",             # Safe storage error normalization
+  "sqlspec/storage/backends/base.py",      # Storage backend runtime base classes (iterators in _iterators.py)
   "sqlspec/data_dictionary/**/*.py",       # Data dictionary mixin (required for adapter inheritance)
   "sqlspec/adapters/**/core.py",           # Adapter compiled helpers
   "sqlspec/adapters/**/type_converter.py", # All adapters type converters

sqlspec/_serialization.py

@@ -11,15 +11,30 @@
 import datetime
 import enum
 import json
+import uuid as _uuid_mod
 from abc import ABC, abstractmethod
 from decimal import Decimal
 from typing import Any, Final, Literal, Protocol, overload
 
-from sqlspec._typing import NUMPY_INSTALLED
+from sqlspec._typing import NUMPY_INSTALLED, UUID_UTILS_INSTALLED
 from sqlspec.core.filters import OffsetPagination
 from sqlspec.typing import MSGSPEC_INSTALLED, ORJSON_INSTALLED, PYDANTIC_INSTALLED, BaseModel
 
 
+def _get_uuid_utils_type() -> "type[Any] | None":
+    if not UUID_UTILS_INSTALLED:
+        return None
+    try:
+        import uuid_utils as _uuid_utils_mod  # pyright: ignore[reportMissingImports]
+    except ImportError:
+        return None
+    else:
+        return _uuid_utils_mod.UUID  # type: ignore[no-any-return,unused-ignore]
+
+
+_UUID_UTILS_TYPE: "type[Any] | None" = _get_uuid_utils_type()
+
+
 def _type_to_string(value: Any) -> Any:  # pragma: no cover
     """Convert special types to strings for JSON serialization.
 
@@ -44,6 +59,10 @@ def _type_to_string(value: Any) -> Any:  # pragma: no cover
         return str(value.value)
     if PYDANTIC_INSTALLED and isinstance(value, BaseModel):
         return value.model_dump_json()
+    if isinstance(value, _uuid_mod.UUID):
+        return str(value)
+    if _UUID_UTILS_TYPE is not None and isinstance(value, _UUID_UTILS_TYPE):
+        return str(value)
     if isinstance(value, OffsetPagination):
         return {"items": value.items, "limit": value.limit, "offset": value.offset, "total": value.total}
     if NUMPY_INSTALLED: