Add tests

R-package/tests/testthat/test-numeric.R

@@ -39,8 +39,8 @@ test_that("NumericSexp works for usize conversions", {
   # f64 to usize
   expect_equal(
     # 2147483647 = .Machine$integer.max
-    usize_to_string(c(0.0, 10.0, 2147483648.0)),
-    c("0", "10", "2147483648")
+    usize_to_string(c(0.0, 10.0, 2147483648.0, 9007199254740991.0)),
+    c("0", "10", "2147483648", "9007199254740991")
   )
 
   # error cases
@@ -50,6 +50,7 @@ test_that("NumericSexp works for usize conversions", {
   expect_error(usize_to_string(NaN))
   expect_error(usize_to_string(-1L))
   expect_error(usize_to_string(-1.0))
+  expect_error(usize_to_string_scalar(9007199254740992.0))
 })
 
 
@@ -81,6 +82,7 @@ test_that("NumericScalar works for usize conversions", {
   expect_equal(usize_to_string_scalar(10.0), "10")
   # 2147483647 = .Machine$integer.max
   expect_equal(usize_to_string_scalar(2147483648.0), "2147483648")
+  expect_equal(usize_to_string_scalar(9007199254740991.0), "9007199254740991")
 
   # error cases
   expect_error(usize_to_string_scalar(NA_integer_))
@@ -89,4 +91,5 @@ test_that("NumericScalar works for usize conversions", {
   expect_error(usize_to_string_scalar(NaN))
   expect_error(usize_to_string_scalar(-1L))
   expect_error(usize_to_string_scalar(-1.0))
+  expect_error(usize_to_string_scalar(9007199254740992.0))
 })

src/sexp/numeric.rs

@@ -8,7 +8,9 @@ const I32MAX: f64 = i32::MAX as f64;
 const I32MIN: f64 = i32::MIN as f64;
 
 // f64 can represent 2^53
-// cf. https://en.wikipedia.org/wiki/Double-precision_floating-point_format
+//
+// cf. https://en.wikipedia.org/wiki/Double-precision_floating-point_format,
+//     https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Number/MAX_SAFE_INTEGER
 const F64_MAX_SIGFIG: f64 = (2_u64.pow(53) - 1) as f64;
 
 const TOLERANCE: f64 = 0.01; // This is super-tolerant than vctrs, but this should be sufficient.