StarMirror/xlnt
1
0
Fork 0
mirror of https://github.com/tfussell/xlnt.git synced 2024-03-22 13:11:17 +08:00
Code Issues Projects Releases Wiki Activity

fix parameter ordering inconsistency that could cause switching the parameters to give different results

Browse Source
This commit is contained in:
Crzyrndm 2018-08-25 13:31:16 +12:00
parent 4188d35caf
commit 875f143d74
2 changed files with 35 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
source/detail/numeric_utils.hpp
View File

@ -88,18 +88,19 @@ bool
float_equals(const LNumber &lhs, const RNumber &rhs,
int epsilon_scale = 20) // scale the "fuzzy" equality. Higher value gives a more tolerant comparison
{
// a type that lhs and rhs can agree on
using common_t = typename std::common_type<LNumber, RNumber>::type;
// asserts for sane usage
static_assert(std::is_floating_point<LNumber>::value || std::is_floating_point<RNumber>::value,
"Using this function with two integers is just wasting time. Use ==");
static_assert(std::is_floating_point<EpsilonType>::value,
"Cannot extract epsilon from a number that isn't a floating point type");
static_assert(std::numeric_limits<EpsilonType>::epsilon() < common_t{1},
"epsilon >= 1.0 will cause all comparisons to return true");
// NANs always compare false with themselves
if (std::isnan(lhs) || std::isnan(rhs))
{
return false;
}
// a type that lhs and rhs can agree on
using common_t = typename std::common_type<LNumber, RNumber>::type;
// epsilon type defaults to float because even if both args are a higher precision type
// either or both could have been promoted by prior operations
// if a higher precision is required, the template type can be changed
@ -108,7 +109,9 @@ float_equals(const LNumber &lhs, const RNumber &rhs,
// epsilon for numeric_limits is valid when abs(x) <1.0, scaling only needs to be upwards
// in particular, this prevents a lhs of 0 from requiring an exact comparison
// additionally, a scale factor is applied.
common_t scaled_fuzz = epsilon_scale * epsilon * max(xlnt::detail::abs<common_t>(lhs), common_t{1});
common_t scaled_fuzz = epsilon_scale * epsilon * max(max(xlnt::detail::abs<common_t>(lhs),
xlnt::detail::abs<common_t>(rhs)), // |max| of parameters.
common_t{1}); // clamp
return ((lhs + scaled_fuzz) >= rhs) && ((rhs + scaled_fuzz) >= lhs);
}

27
tests/detail/numeric_util_test_suite.cpp
View File

@ -31,6 +31,7 @@ public:
{
register_test(test_float_equals_zero);
register_test(test_float_equals_large);
register_test(test_float_equals_fairness);
register_test(test_min);
register_test(test_max);
register_test(test_abs);
@ -121,6 +122,32 @@ public:
xlnt_assert(!xlnt::detail::float_equals(nan, 1000.f));
}
void test_float_equals_fairness()
{
// tests for parameter ordering dependency
// (lhs ~= rhs) == (rhs ~= lhs)
const double test_val = 1.0;
const double test_diff_pass = 1.192092e-07; // should all pass with this
const double test_diff = 1.192093e-07; // difference enough to provide different results if the comparison is not "fair"
const double test_diff_fails = 1.192094e-07; // should all fail with this
// test_diff_pass
xlnt_assert(xlnt::detail::float_equals<float>((test_val + test_diff_pass), test_val, 1));
xlnt_assert(xlnt::detail::float_equals<float>(test_val, (test_val + test_diff_pass), 1));
xlnt_assert(xlnt::detail::float_equals<float>(-(test_val + test_diff_pass), -test_val, 1));
xlnt_assert(xlnt::detail::float_equals<float>(-test_val, -(test_val + test_diff_pass), 1));
// test_diff
xlnt_assert(xlnt::detail::float_equals<float>((test_val + test_diff), test_val, 1));
xlnt_assert(xlnt::detail::float_equals<float>(test_val, (test_val + test_diff), 1));
xlnt_assert(xlnt::detail::float_equals<float>(-(test_val + test_diff), -test_val, 1));
xlnt_assert(xlnt::detail::float_equals<float>(-test_val, -(test_val + test_diff), 1));
// test_diff_fails
xlnt_assert(!xlnt::detail::float_equals<float>((test_val + test_diff_fails), test_val, 1));
xlnt_assert(!xlnt::detail::float_equals<float>(test_val, (test_val + test_diff_fails), 1));
xlnt_assert(!xlnt::detail::float_equals<float>(-(test_val + test_diff_fails), -test_val, 1));
xlnt_assert(!xlnt::detail::float_equals<float>(-test_val, -(test_val + test_diff_fails), 1));
}
void test_min()
{
// simple