⚡️ Speed up method _AppConfig.asdict by 827%

[codeflash-ai]

📄 827% (8.27x) speedup for _AppConfig.asdict in marimo/_ast/app_config.py

⏱️ Runtime : 5.81 milliseconds → 627 microseconds (best of 250 runs)

📝 Explanation and details

The optimization replaces the expensive asdict(self) call with direct attribute access through getattr(self, k) while iterating over self.__dataclass_fields__.

Key Performance Changes:

• Eliminates intermediate dictionary creation: The original code calls dataclasses.asdict(self) which creates a complete dictionary of all fields, then filters it. The optimized version directly accesses only the needed attributes.
• Reduces memory allocations: Instead of creating a full dictionary and then filtering it, the optimized code builds only the final result dictionary.
• Avoids unnecessary field processing: dataclasses.asdict() performs additional work like handling nested dataclasses and type conversions that aren't needed here.

Why This Works:
The line profiler shows the optimization dramatically reduces time spent in the dictionary comprehension (from 89.6% to 3.6% of total time). The asdict(self).items() call in the original code is expensive because it processes all dataclass fields, even those that will be filtered out.

Test Results Impact:
The annotated tests show consistent 5-6x speedups across all scenarios:

• Simple default configs: ~500-600% faster
• Configs with private fields: ~550% faster
• Configs with many private fields: ~500% faster

This optimization is particularly effective for dataclasses with many fields or when called frequently, as it eliminates the overhead of the full dataclass serialization process while maintaining identical functionality.

✅ Correctness verification report:

Test	Status
⚙️ Existing Unit Tests	✅ 26 Passed
🌀 Generated Regression Tests	✅ 12 Passed
⏪ Replay Tests	🔘 None Found
🔎 Concolic Coverage Tests	🔘 None Found
📊 Tests Coverage	100.0%

⚙️ Existing Unit Tests and Runtime

Test File::Test Function	Original ⏱️	Optimized ⏱️	Speedup
_ast/test_app.py::TestApp.test_app_config_auto_download	8.54μs	2.13μs	300%✅
_ast/test_app.py::TestApp.test_app_config_extra_args_ignored	16.2μs	3.16μs	414%✅
_ast/test_app.py::test_app_config	13.9μs	2.27μs	512%✅
_ast/test_app.py::test_app_config_extra_args_ignored	14.9μs	2.49μs	499%✅
_ast/test_app_config.py::test_app_config_asdict	15.5μs	2.30μs	571%✅

🌀 Generated Regression Tests and Runtime

from dataclasses import dataclass, field
from typing import Any, Optional

imports

import pytest
from marimo._ast.app_config import _AppConfig

---- Mocking dependencies for a self-contained test ----

These are minimal mocks to allow the test to run without external dependencies.

Simulate valid types for the config fields

WidthType = str # could be "compact", "full", etc.
ExportType = str # could be "html", "md", etc.
SqlOutputType = str # could be "auto", "table", etc.
from marimo._ast.app_config import _AppConfig

-------------------- UNIT TESTS --------------------

1. Basic Test Cases

def test_default_config_asdict():
"""Test asdict() on default configuration."""
cfg = _AppConfig()
codeflash_output = cfg.asdict(); d = codeflash_output # 13.6μs -> 2.23μs (509% faster)

def test_auto_download_empty_and_nonempty():
"""Test asdict() with empty and non-empty auto_download lists."""
# Empty list (default)
cfg = _AppConfig()
codeflash_output = cfg.asdict(); d = codeflash_output # 13.3μs -> 2.15μs (519% faster)
# Non-empty list
cfg2 = _AppConfig(auto_download=["md"])
codeflash_output = cfg2.asdict(); d2 = codeflash_output # 8.88μs -> 1.22μs (626% faster)

def test_asdict_return_type_and_keys():
"""Test that asdict returns a dict with correct keys and types."""
cfg = _AppConfig()
codeflash_output = cfg.asdict(); d = codeflash_output # 16.3μs -> 2.56μs (535% faster)
expected_keys = [
"width", "app_title", "layout_file", "css_file", "html_head_file", "auto_download", "sql_output"
]

codeflash_output is used to check that the output of the original code is the same as that of the optimized code.

#------------------------------------------------
from dataclasses import asdict, dataclass, field
from typing import Any, Optional

imports

import pytest
from marimo._ast.app_config import _AppConfig

Dummy enum types to simulate the real ones for testing

class ExportType(str):
HTML = "html"
MARKDOWN = "markdown"
PDF = "pdf"

class SqlOutputType(str):
AUTO = "auto"
TABLE = "table"
RAW = "raw"

class WidthType(str):
COMPACT = "compact"
FULL = "full"
CUSTOM = "custom"
from marimo._ast.app_config import _AppConfig

unit tests

----------- Basic Test Cases -----------

def test_asdict_default_values():
"""Test asdict with all default values."""
cfg = _AppConfig()
codeflash_output = cfg.asdict(); d = codeflash_output # 14.5μs -> 2.37μs (511% faster)

def test_asdict_ignores_private_fields():
"""Test that asdict does not include private fields (starting with '_')."""
# Dynamically add a private field to the instance
cfg = _AppConfig()
cfg._experimental = 123
# asdict() should not include '_experimental'
codeflash_output = cfg.asdict(); d = codeflash_output # 12.7μs -> 1.93μs (557% faster)

def test_asdict_large_number_of_fields_with_private():
"""Test asdict ignores many private fields even if present."""
cfg = _AppConfig()
# Add many private fields
for i in range(100):
setattr(cfg, f"_private{i}", i)
codeflash_output = cfg.asdict(); d = codeflash_output # 14.3μs -> 2.39μs (497% faster)
# None of the private fields should be present
for i in range(100):
pass

codeflash_output is used to check that the output of the original code is the same as that of the optimized code.

To edit these changes git checkout codeflash/optimize-_AppConfig.asdict-mhvkqyeo and push.