René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Invalid corner cases (resulting in nan+nanj) in complex division","articleBody":"# Bug report\r\n\r\n### Bug description:\r\n\r\nReproducer:\r\n```pycon\r\n\u003e\u003e\u003e (1+1j)/(cmath.inf+0j)  # ok\r\n0j\r\n\u003e\u003e\u003e (1+1j)/cmath.infj  # ok\r\n-0j\r\n\u003e\u003e\u003e (1+1j)/(cmath.inf+cmath.infj)  # should be 0j, isn't?\r\n(nan+nanj)\r\n```\r\nc.f. MPC:\r\n```pycon\r\n\u003e\u003e\u003e gmpy2.mpc('(1+1j)')/gmpy2.mpc('(inf+infj)')\r\nmpc('0.0+0.0j')\r\n```\r\n\r\nIt seems, there are not so much such cases (i.e. when the numerator is finite and the denominator has infinite and infinite/nan components, or if the numerator has infinite components and the denominator has zeros).  Following patch (mostly literally taken from the C11 Annex G.5.1 ``_Cdivd()``'s example, except for the ``denom == 0.0`` case) should fix the issue:\r\n```diff\r\ndiff --git a/Objects/complexobject.c b/Objects/complexobject.c\r\nindex b3d57b8337..9041cbb8ae 100644\r\n--- a/Objects/complexobject.c\r\n+++ b/Objects/complexobject.c\r\n@@ -122,6 +122,27 @@ _Py_c_quot(Py_complex a, Py_complex b)\r\n         /* At least one of b.real or b.imag is a NaN */\r\n         r.real = r.imag = Py_NAN;\r\n     }\r\n+\r\n+    /* Recover infinities and zeros that computed as nan+nanj */\r\n+    if (isnan(r.real) \u0026\u0026 isnan(r.imag)) {\r\n+        if ((isinf(a.real) || isinf(a.imag))\r\n+            \u0026\u0026 isfinite(b.real) \u0026\u0026 isfinite(b.imag))\r\n+        {\r\n+            const double x = copysign(isinf(a.real) ? 1.0 : 0.0, a.real);\r\n+            const double y = copysign(isinf(a.imag) ? 1.0 : 0.0, a.imag);\r\n+            r.real = Py_INFINITY * (x*b.real + y*b.imag);\r\n+            r.imag = Py_INFINITY * (y*b.real - x*b.imag);\r\n+        }\r\n+        else if ((fmax(abs_breal, abs_bimag) == Py_INFINITY)\r\n+                 \u0026\u0026 isfinite(a.real) \u0026\u0026 isfinite(a.imag))\r\n+        {\r\n+            const double x = copysign(isinf(b.real) ? 1.0 : 0.0, b.real);\r\n+            const double y = copysign(isinf(b.imag) ? 1.0 : 0.0, b.imag);\r\n+            r.real = 0.0 * (a.real*x + a.imag*y);\r\n+            r.imag = 0.0 * (a.imag*x - a.real*y);\r\n+        }\r\n+    }\r\n+\r\n     return r;\r\n }\r\n \r\n```\r\n\r\nPerhaps, we could also clear ending remark in ``_Py_c_quot()`` comment:\r\nhttps://github.com/python/cpython/blob/d065edfb66470bbf06367b3570661d0346aa6707/Objects/complexobject.c#L91-L92\r\n\r\nI'll prepare a PR, if this issue will be accepted.\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\n_No response_\n\n\u003c!-- gh-linked-prs --\u003e\n### Linked PRs\n* gh-119457\n\u003c!-- /gh-linked-prs --\u003e\n","author":{"url":"https://github.com/skirpichev","@type":"Person","name":"skirpichev"},"datePublished":"2024-05-22T02:07:17.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":4},"url":"https://github.com/119372/cpython/issues/119372"}