Test failures on 32bit systems #310
Description
It looks like v1.9.0 increased test coverage, and on 32bit systems (like armhf), the int and uint tests fail because those types are too small to hold some of the values:
--- FAIL: TestNumber/int64/#14/Value/Pointer/ToE (0.01s)
number_test.go:381:
error:
got non-nil error
got:
e`unable to cast "9223372036854775807" of type string to int64: strconv.ParseInt: parsing "9223372036854775807": value out of range`
stack:
/home/gibmat/cast/number_test.go:381
c.Assert(err, qt.IsNil)
--- FAIL: TestNumber/uint64/#14/Value/Pointer/ToTypeE (0.01s)
number_test.go:367:
error:
got non-nil error
got:
e`unable to cast "18446744073709551615" of type string to uint64: strconv.ParseUint: parsing "18446744073709551615": value out of range`
stack:
/home/gibmat/cast/number_test.go:367
c.Assert(err, qt.IsNil)
I looked at the code, and explicitly telling strconv.Parse{Int,Uint} to use a full 64bit variable allows the tests to pass on both 32bit and 64bit systems. However, I don't know if this might have other consequence if trying to cast an 8, 16, or 32bit variable.
diff --git a/number.go b/number.go
index a58dc4d..a4c4576 100644
--- a/number.go
+++ b/number.go
@@ -405,7 +405,7 @@ func parseNumber[T Number](s string) (T, error) {
}
func parseInt[T integer](s string) (T, error) {
- v, err := strconv.ParseInt(trimDecimal(s), 0, 0)
+ v, err := strconv.ParseInt(trimDecimal(s), 0, 64)
if err != nil {
return 0, err
}
@@ -414,7 +414,7 @@ func parseInt[T integer](s string) (T, error) {
}
func parseUint[T unsigned](s string) (T, error) {
- v, err := strconv.ParseUint(strings.TrimLeft(trimDecimal(s), "+"), 0, 0)
+ v, err := strconv.ParseUint(strings.TrimLeft(trimDecimal(s), "+"), 0, 64)
if err != nil {
return 0, err
}