René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"CTypes test failed, complex double problem","articleBody":"# Bug report\r\n\r\n### Bug description:\r\n\r\nPython binary has been built in the standard way:\r\n```\r\n./configure --with-pydebug \u0026\u0026 make -j\r\n```\r\n\r\nBut test_ctypes fails:\r\n```\r\n-\u003e % ./python -m unittest -v test.test_ctypes.test_libc.LibTest.test_csqrt \r\ntest_csqrt (test.test_ctypes.test_libc.LibTest.test_csqrt) ... FAIL\r\n\r\n======================================================================\r\nFAIL: test_csqrt (test.test_ctypes.test_libc.LibTest.test_csqrt)\r\n----------------------------------------------------------------------\r\nTraceback (most recent call last):\r\n  File \"/home/mikhail.efimov/projects/cpython/Lib/test/test_ctypes/test_libc.py\", line 30, in test_csqrt\r\n    self.assertEqual(lib.my_csqrt(4), 2+0j)\r\n    ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^\r\nAssertionError: (5e-324+6.95324111713477e-310j) != (2+0j)\r\n\r\n----------------------------------------------------------------------\r\nRan 1 test in 0.001s\r\n\r\nFAILED (failures=1)\r\n```\r\n\r\nSystem:\r\n```\r\n-\u003e % cat /etc/issue\r\nDebian GNU/Linux 10 \\n \\l\r\n```\r\n\r\nGCC:\r\n```\r\n-\u003e % gcc -v\r\nUsing built-in specs.\r\nCOLLECT_GCC=gcc\r\nCOLLECT_LTO_WRAPPER=/usr/lib/gcc/x86_64-linux-gnu/8/lto-wrapper\r\nOFFLOAD_TARGET_NAMES=nvptx-none\r\nOFFLOAD_TARGET_DEFAULT=1\r\nTarget: x86_64-linux-gnu\r\nConfigured with: ../src/configure -v --with-pkgversion='Debian 8.3.0-6' --with-bugurl=file:///usr/share/doc/gcc-8/README.Bugs --enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++ --prefix=/usr --with-gcc-major-version-only --program-suffix=-8 --program-prefix=x86_64-linux-gnu- --enable-shared --enable-linker-build-id --libexecdir=/usr/lib --without-included-gettext --enable-threads=posix --libdir=/usr/lib --enable-nls --enable-bootstrap --enable-clocale=gnu --enable-libstdcxx-debug --enable-libstdcxx-time=yes --with-default-libstdcxx-abi=new --enable-gnu-unique-object --disable-vtable-verify --enable-libmpx --enable-plugin --enable-default-pie --with-system-zlib --with-target-system-zlib --enable-objc-gc=auto --enable-multiarch --disable-werror --with-arch-32=i686 --with-abi=m64 --with-multilib-list=m32,m64,mx32 --enable-multilib --with-tune=generic --enable-offload-targets=nvptx-none --without-cuda-driver --enable-checking=release --build=x86_64-linux-gnu --host=x86_64-linux-gnu --target=x86_64-linux-gnu\r\nThread model: posix\r\ngcc version 8.3.0 (Debian 8.3.0-6) \r\n```\r\n\r\nI've tried to fix it by myself but the result has not been achieved in a reasonable amount of time.\r\nThere is a simple test I've provided:\r\n```\r\n-\u003e % cat test_csqrt.c\r\n#include \u003ccomplex.h\u003e\r\n#include \u003cstdio.h\u003e\r\n\r\nint complex my_csqrt(double complex a)\r\n{\r\n    double complex z1 = a;\r\n    double complex z2 = csqrt(a);\r\n    printf(\"my_csqrt (%.10f%+.10fi) = %.10f%+.10fi\\n\",\r\n        creal(z1), cimag(z1), creal(z2), cimag(z2));\r\n    return 0;\r\n}\r\n\r\nint main() {\r\n    my_csqrt(4.0);\r\n    my_csqrt(4.0+4.0j);\r\n    my_csqrt(-1+0.01j);\r\n    my_csqrt(-1-0.01j);\r\n    return 0;\r\n}\r\n```\r\n\r\n```\r\n-\u003e % gcc -lm test_csqrt.c -o test_csqrt \u0026\u0026 ./test_csqrt\r\nmy_csqrt (4.0000000000+0.0000000000i) = 2.0000000000+0.0000000000i\r\nmy_csqrt (4.0000000000+4.0000000000i) = 2.1973682269+0.9101797211i\r\nmy_csqrt (-1.0000000000+0.0100000000i) = 0.0049999375+1.0000124996i\r\nmy_csqrt (-1.0000000000-0.0100000000i) = 0.0049999375-1.0000124996i\r\n```\r\n\r\nSo, it's not a problem in my libc version.\r\n\r\nMoreover, this problem can be reproduced with standard libm.so (/lib/x86_64-linux-gnu/libm-2.28.so):\r\n```\r\n-\u003e % cat ctypes_fix/test.py      \r\nimport ctypes\r\nlibm = ctypes.CDLL('libm.so.6')\r\nlibm.clog.restype = ctypes.c_double_complex\r\nlibm.clog.argtypes = ctypes.c_double_complex,\r\nclog_5 = libm.clog(5.0)\r\nclog_1000_2j = libm.clog(1000.0+2j)\r\nprint(f\"{clog_5=}\")\r\nprint(f\"{clog_1000_2j=}\")\r\n```\r\n```\r\n-\u003e % ./python ctypes_fix/test.py\r\nclog_5=(5e-324+6.9529453382261e-310j)\r\nclog_1000_2j=(5e-324+6.9529453382261e-310j)\r\n```\r\n\r\nIMHO, some problem lies in using ctypes.c_double_complex as an argument and return value types.\r\nFYI, with double argtype and restype clog works like classical double log:\r\n\r\n```\r\n-\u003e % cat ctypes_fix/test2.py   \r\nimport ctypes\r\nlibm = ctypes.CDLL('libm.so.6')\r\nlibm.clog.restype = ctypes.c_double\r\nlibm.clog.argtypes = ctypes.c_double,\r\nclog_5 = libm.clog(5.0)\r\nclog_1000 = libm.clog(1000.0)\r\nprint(f\"{clog_5=}\")\r\nprint(f\"{clog_1000=}\")\r\n```\r\n```\r\n-\u003e % ./python ctypes_fix/test2.py\r\nclog_5=1.6094379124341003\r\nclog_1000=6.907755278982137\r\n```\r\n\r\n### CPython versions tested on:\r\n\r\nCPython main branch\r\n\r\n### Operating systems tested on:\r\n\r\nLinux\n\n\u003c!-- gh-linked-prs --\u003e\n### Linked PRs\n* gh-125322\n* gh-126104\n* gh-132865\n* gh-135932\n* gh-135973\n\u003c!-- /gh-linked-prs --\u003e\n","author":{"url":"https://github.com/efimov-mikhail","@type":"Person","name":"efimov-mikhail"},"datePublished":"2024-10-09T18:34:42.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":10},"url":"https://github.com/125206/cpython/issues/125206"}