gh-103000: Optimise `dataclasses.asdict` for the common case by AlexWaygood · Pull Request #104364 · python/cpython

Now that PEP 709 has been implemented, this change speeds up the following benchmark by 15% on my machine (Windows, PGO-optimised non-debug build):

from dataclasses import dataclass, asdict
import timeit

@dataclass
class Foo:
    x: int

@dataclass
class Bar:
    x: Foo
    y: Foo
    z: Foo

@dataclass
class Baz:
    x: Bar
    y: Bar
    z: Bar

foo = Foo(42)
bar = Bar(foo, foo, foo)
baz = Baz(bar, bar, bar)

print(timeit.timeit(lambda: asdict(baz), number=250_000))

This idea was originally @DavidCEllis's, but prior to PEP 709, comprehensions had too much overhead to make it worthwhile. Now that PEP 709 has been implemented, the impact of the patch seems pretty different.

Co-authored-by: David Ellis ducksual@gmail.com

• Issue: Dataclasses - Improve the performance of asdict/astuple for common types and default values #103000