fix(truncate): add machine-consistent behaviour for undefined NaN error

Closes #280

caput/truncate.pyx

@@ -64,7 +64,14 @@ def bit_truncate_float(float val, float err):
     -------
     val
         The truncated value.
+
+    Raises
+    ------
+    ValueError
+        If `err` is a NaN.
     """
+    if err != err:
+        raise ValueError(f"Error {err} is invalid.")
 
     return _bit_truncate_float(val, err)
 
@@ -83,7 +90,14 @@ def bit_truncate_double(double val, double err):
     -------
     val
         The truncated value.
+
+    Raises
+    ------
+    ValueError
+        If `err` is a NaN.
     """
+    if err != err:
+        raise ValueError(f"Error {err} is invalid.")
 
     return _bit_truncate_double(val, err)
 

tests/test_truncate.py

@@ -1,4 +1,5 @@
 import numpy as np
+import pytest
 
 from caput import truncate
 
@@ -58,21 +59,23 @@ def test_truncate_float():
     assert truncate.bit_truncate_double(32.121, -1) == 0
     assert truncate.bit_truncate_float(32.121, np.inf) == 0
     assert truncate.bit_truncate_double(32.121, np.inf) == 0
-    assert truncate.bit_truncate_float(32.121, np.nan) == 0
-    assert truncate.bit_truncate_double(32.121, np.nan) == 0
     assert truncate.bit_truncate_float(np.inf, 1) == np.inf
     assert truncate.bit_truncate_double(np.inf, 1) == np.inf
     assert np.isnan(truncate.bit_truncate_float(np.nan, 1))
     assert np.isnan(truncate.bit_truncate_double(np.nan, 1))
+
     assert truncate.bit_truncate_float(np.inf, np.inf) == 0
     assert truncate.bit_truncate_double(np.inf, np.inf) == 0
-    assert truncate.bit_truncate_float(np.inf, np.nan) == 0
-    assert truncate.bit_truncate_double(np.inf, np.nan) == 0
-    assert truncate.bit_truncate_float(np.nan, np.nan) == 0
-    assert truncate.bit_truncate_double(np.nan, np.nan) == 0
     assert truncate.bit_truncate_float(np.nan, np.inf) == 0
     assert truncate.bit_truncate_double(np.nan, np.inf) == 0
 
+    # Test that an error is raised when `err` is `NaN`
+    with pytest.raises(ValueError):
+        truncate.bit_truncate_float(32.121, np.nan)
+
+    with pytest.raises(ValueError):
+        truncate.bit_truncate_double(32.121, np.nan)
+
 
 def test_truncate_array():
     assert (