Remove the test that ensures that when the string "nan(...)" is converted

to floating-point, the result is a quiet NaN.  The current implementation
may return a signaling NaN, and the vendor has no plans for changing this,
for reasons explained in the comment I added.

1 files changed, 8 insertions, 0 deletions

diff --git a/tools/regression/lib/libc/stdio/test-scanfloat.c b/tools/regression/lib/libc/stdio/test-scanfloat.c
index 342fc73..ef6faae 100644
--- a/tools/regression/lib/libc/stdio/test-scanfloat.c
+++ b/tools/regression/lib/libc/stdio/test-scanfloat.c
@@ -190,7 +190,15 @@ main(int argc, char *argv[])
 	assert(f != f);
 	assert(d != d);
 	assert(ld != ld);
+#if 0
+	/*
+	 * POSIX says we should only generate quiet NaNs, but the gdtoa
+	 * author convincingly argues that if you ask for a NaN format
+	 * based on some implementation-defined string, you should get
+	 * what you asked for, even if it's a signaling NaN.
+	 */
 	assert(fetestexcept(FE_INVALID) == 0);
+#endif
 
 	printf("ok 2 - scanfloat\n");