May 26, 2025 · Apr 29, 2025 · Apr 29, 2025 · Apr 29, 2025 · Apr 29, 2025 · Apr 30, 2025
diff --git a/Misc/NEWS.d/next/Library/2025-04-29-11-48-46.gh-issue-132876.lyTQGZ.rst b/Misc/NEWS.d/next/Library/2025-04-29-11-48-46.gh-issue-132876.lyTQGZ.rst
 Workaround broken ``ldexp()`` implementation on Windows 10.
 `ldexp()` on Windows doesn't round subnormal results before Windows 11, but should. Python's :func:`math.ldexp` wrapper now does round them, so results may change
 slightly, in rare cases of very small results, on Windows versions before 11.
diff --git a/Modules/mathmodule.c b/Modules/mathmodule.c
        errno = 0;
        r = ldexp(x, (int)exp);
 #ifdef _MSC_VER
        if (-DBL_MIN< r && r<DBL_MIN) {
        if (DBL_MIN> r && r> -DBL_MIN) {
            /* Denormal (or zero) results can be incorrectly rounded here (rather,
               truncated).  Fixed in newer versions of the C runtime, included
               with Windows 11. */
Original file line number	Diff line number	Diff line change
		@@ -1 +1,2 @@
		Workaround broken ``ldexp()`` implementation on Windows 10.
		`ldexp()` on Windows doesn't round subnormal results before Windows 11, but should. Python's :func:`math.ldexp` wrapper now does round them, so results may change
		slightly, in rare cases of very small results, on Windows versions before 11.
Original file line number	Diff line number	Diff line change
Expand Up		@@ -2164,7 +2164,7 @@ math_ldexp_impl(PyObject module, double x, PyObject i)
		errno = 0;
		r = ldexp(x, (int)exp);
		#ifdef _MSC_VER
		if (-DBL_MIN< r && r<DBL_MIN) {
		if (DBL_MIN> r && r> -DBL_MIN) {
		/* Denormal (or zero) results can be incorrectly rounded here (rather,
		truncated). Fixed in newer versions of the C runtime, included
		with Windows 11. */
Expand Down