amorfo77/json
1
0 Fork 0
mirror of https://github.com/nlohmann/json.git synced 2026-10-22 22:08:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

Reduce test compile times (spike)

Browse Source 
Two changes roughly halve the CPU time needed to build the unit tests
(clang 21: ~300 s -> ~150 s, GCC 16: ~750-820 s -> ~395-420 s, Debug):

- Move the C++14/17/20-dependent tests into separate
  unit-<name>-cpp<N>.cpp files. A test file is built for every standard
  whose JSON_HAS_CPP_<N> macro it mentions, so far whole large files were
  rebuilt for C++14/17/20 because of a few #ifdef sections. The main files
  are now built for C++11 only; the ci_test_*_cxx<N> jobs still build
  every file for every standard.

- Add the CMake option JSON_TestUnityBuild (ON by default, OFF with
  MinGW): compatible test files are compiled in batches as one
  translation unit so they share the template instantiations of the
  library. Each file keeps its own CTest test, which runs the batch
  executable filtered to that file's test cases. The binary-format tests
  form an explicit group; the rest is batched by JSON_TestUnityBatchSize.

Fix the name clashes that merging files exposed, document the rules for
test files in tests/README.md, and point CONTRIBUTING.md to it.

Signed-off-by: Niels Lohmann <mail@nlohmann.me>
This commit is contained in:
Niels Lohmann committed 2026-10-09 16:49:45 +02:00
1 parent d33068da73
commit 993e37e7ec
44 files changed
+3965 -2225
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

tests/src/unit-deserialization.cpp
+1 -36
View File
@@ -1306,44 +1306,9 @@ TEST_CASE("deserialization")
CHECK(line == "foo");
}
}
// build with C++20
// JSON_HAS_CPP_20
#if defined(__cpp_char8_t)
SECTION("Using _json with char8_t literals #4945")
{
// Regular narrow string literal
const auto j1 = R"({"key": "value", "num": 42})"_json;
CHECK(j1["key"] == "value");
CHECK(j1["num"] == 42);
// UTF-8 prefixed literal (C++20 and later); the emoji is written as a
// \U escape rather than a raw multibyte character so this does not
// depend on the compiler's source-file encoding (e.g., MSVC without
// /utf-8, or classic ICC, which does not encode non-ASCII narrow
// string literals as UTF-8 - compare against a \x-escaped expectation
// for the same reason)
const auto j2 = u8"{\"emoji\": \"\U0001F600\", \"msg\": \"hello\"}"_json;
CHECK(j2["emoji"] == "\xF0\x9F\x98\x80");
CHECK(j2["msg"] == "hello");
const auto j3 = u8R"({"key": "value", "num": 42})"_json;
CHECK(j3["key"] == "value");
CHECK(j3["num"] == 42);
}
#endif
}
// select the types to test - char8_t is only available since C++20 if and only
// if __cpp_char8_t is defined.
#define TYPE_LIST(...) __VA_ARGS__
#if defined(__cpp_char8_t) && (__cpp_char8_t >= 201811L)
#define ASCII_TYPES TYPE_LIST(char, wchar_t, char16_t, char32_t, char8_t)
#else
#define ASCII_TYPES TYPE_LIST(char, wchar_t, char16_t, char32_t)
#endif
TEST_CASE_TEMPLATE("deserialization of different character types (ASCII)", T, ASCII_TYPES) // NOLINT(readability-math-missing-parentheses, bugprone-throwing-static-initialization)
TEST_CASE_TEMPLATE("deserialization of different character types (ASCII)", T, char, wchar_t, char16_t, char32_t) // NOLINT(readability-math-missing-parentheses, bugprone-throwing-static-initialization)
{
std::vector<T> const v = {'t', 'r', 'u', 'e'};
CHECK(json::parse(v) == json(true));