Add float16 tests

Author: Bisaloo

inst/extdata/test_float16.npy

@@ -14,6 +14,7 @@
 np.save("inst/extdata/test_uint64_overflowing.npy", np.array([0, 1, 10, 100, 1000, 10000, 100000, 2**16, 2**32, 2**48, 2**63, 2**64 - 1], dtype="uint64"))
 
 # Float types
+np.save("inst/extdata/test_float16.npy", np.array([1.5, -2.5, 3.14, 0.0, -1.0, 100.0, -100.0, 0.001, 1000, -1000, 1.2345, -9.8765], dtype="float16"))
 np.save("inst/extdata/test_float32.npy", np.array([1.5, -2.5, 3.14, 0.0, -1.0, 100.0, -100.0, 0.001, 1e6, -1e6, 1.23456, -9.87654], dtype="float32"))
 np.save("inst/extdata/test_float64.npy", np.array([1.5, -2.5, 3.14, 0.0, -1.0, 100.0, -100.0, 0.001, 1e6, -1e6, 1.23456789, -9.87654321], dtype="float64"))
 

inst/scripts/generate_test_data.py

@@ -109,6 +109,30 @@ test_that("different types work", {
     )) |>
     expect_warning("overflow")
 
+  system.file("extdata", "test_float16.npy", package = "grumpy") |>
+    read_npy() |>
+    expect_no_condition() |>
+    expect_equal(
+      array(
+        c(
+          1.5,
+          -2.5,
+          3.14,
+          0.0,
+          -1.0,
+          100.0,
+          -100.0,
+          0.001,
+          1000.0,
+          -1000.0,
+          1.2345,
+          -9.8765
+        ),
+        dim = 12L
+      ),
+      tolerance = 1e-3
+    )
+
   system.file("extdata", "test_float32.npy", package = "grumpy") |>
     read_npy() |>
     expect_no_condition() |>