nixos/nix
1
1
Fork 0
mirror of https://github.com/NixOS/nix.git synced 2025-11-27 04:30:59 +01:00
Code Activity

Minimize the use of C Macros for characterization tests

Browse source 
Fewer macros is better!

Introduce a new `JsonChacterizationTest` mixin class to help with this.

Also, avoid some needless copies with `GetParam`.

Part of my effort shoring up the JSON formats with #13570.
This commit is contained in:
John Ericson 2025-09-24 11:55:21 -04:00
parent 3bf1268ac6
commit 01b2037bc0
15 changed files with 364 additions and 252 deletions
Download patch file Download diff file Expand all files Collapse all files

12
src/libutil-tests/sort.cc
View file

@ -102,14 +102,14 @@ struct RandomPeekSort : public ::testing::TestWithParam<
void SetUp() override
{
auto [maxSize, min, max, iterations] = GetParam();
const auto & [maxSize, min, max, iterations] = GetParam();
urng_ = std::mt19937(GTEST_FLAG_GET(random_seed));
distribution_ = std::uniform_int_distribution<int>(min, max);
}
auto regenerate()
{
auto [maxSize, min, max, iterations] = GetParam();
const auto & [maxSize, min, max, iterations] = GetParam();
std::size_t dataSize = std::uniform_int_distribution<std::size_t>(0, maxSize)(urng_);
data_.resize(dataSize);
std::generate(data_.begin(), data_.end(), [&]() { return distribution_(urng_); });
@ -118,7 +118,7 @@ struct RandomPeekSort : public ::testing::TestWithParam<
TEST_P(RandomPeekSort, defaultComparator)
{
auto [maxSize, min, max, iterations] = GetParam();
const auto & [maxSize, min, max, iterations] = GetParam();
for (std::size_t i = 0; i < iterations; ++i) {
regenerate();
@ -132,7 +132,7 @@ TEST_P(RandomPeekSort, defaultComparator)
TEST_P(RandomPeekSort, greater)
{
auto [maxSize, min, max, iterations] = GetParam();
const auto & [maxSize, min, max, iterations] = GetParam();
for (std::size_t i = 0; i < iterations; ++i) {
regenerate();
@ -146,7 +146,7 @@ TEST_P(RandomPeekSort, greater)
TEST_P(RandomPeekSort, brokenComparator)
{
auto [maxSize, min, max, iterations] = GetParam();
const auto & [maxSize, min, max, iterations] = GetParam();
/* This is a pretty nice way of modeling a worst-case scenario for a broken comparator.
If the sorting algorithm doesn't break in such case, then surely all deterministic
@ -170,7 +170,7 @@ TEST_P(RandomPeekSort, brokenComparator)
TEST_P(RandomPeekSort, stability)
{
auto [maxSize, min, max, iterations] = GetParam();
const auto & [maxSize, min, max, iterations] = GetParam();
for (std::size_t i = 0; i < iterations; ++i) {
regenerate();