C Uzantıları Neden Kullanılır? Kullanım Alanları ve Ödünleşimler

C Uzantısı Nedir

# After building the extension:
import myext
result = myext.fast_sum([1, 2, 3, 4, 5])
print(result)   # 15 (computed in C)

C Uzantıları Ne Zaman Kullanılmalı

# Good candidates for C extensions:
# - NumPy array operations
# - Image processing pixel loops
# - Cryptographic primitives
# - Wrapping libpng, libssl, sqlite3

C Uzantılarına Alternatifler

# Cython example — almost Python syntax:
# def fast_sum(int[:] arr) -> int:
#     cdef int total = 0
#     for x in arr:
#         total += x
#     return total

GIL Etkisi

# In a C extension:
# Py_BEGIN_ALLOW_THREADS
#     result = heavy_cpu_work(data);
# Py_END_ALLOW_THREADS
#
# Other Python threads run while this C code executes

Ödünleşimler

# Trade-offs:
# + Fastest possible execution
# + Direct access to C libraries
# + Can release GIL
# - Complex build setup (setup.py / CMake)
# - Platform-specific compilation
# - Memory management is manual
# - Harder to debug

Python/C API'sine Genel Bakış

// C API basics:
// PyObject*  — pointer to any Python object
// Py_INCREF  — increment reference count
// Py_DECREF  — decrement reference count
// PyArg_ParseTuple — parse Python args from C
// PyErr_SetString  — raise a Python exception from C

setup.py ile Derleme

# setup.py
from setuptools import setup, Extension

setup(
    name="myext",
    ext_modules=[
        Extension("myext", sources=["myext.c"])
    ]
)

# Build:
# python setup.py build_ext --inplace

Gerçek Dünya Örnekleri

# NumPy is a C extension:
import numpy as np
a = np.arange(1_000_000)
result = a.sum()   # executed in C, 100x faster than pure Python
print(result)

Cython: Daha Kolay Yol

# fib.pyx (Cython)
cpdef long fib(int n):
    if n < 2:
        return n
    return fib(n-1) + fib(n-2)

# Compile:
# cythonize -i fib.pyx
# import fib; fib.fib(40)

C Uzantılarında Hata Ayıklama

# Compile with debug symbols:
# CFLAGS="-g -O0" python setup.py build_ext --inplace

# Debug:
# gdb python
# (gdb) run script.py
# (gdb) backtrace

# Valgrind for memory leaks:
# valgrind --tool=memcheck python script.py

C Uzantılarını Test Etme

# test_myext.py
import pytest
import myext

def test_fast_sum():
    assert myext.fast_sum([1, 2, 3]) == 6

def test_empty():
    assert myext.fast_sum([]) == 0

def test_type_error():
    with pytest.raises(TypeError):
        myext.fast_sum("not a list")

Hızlı Kontrol

Özet