So that global test environments are by default set up and torn down once,
regardless of the value of the repeat flag.
The point of global environments is to be set up and torn down once, and shared
by all tests in the process. There is no obvious reason why multiple runs of the
same test should be treated distinctly from single runs of different tests.
Having this be false by default means that repeats using a global environment
run faster. It can still be set to true if it's desired that every repeat get a
fresh environment, but this seems less important given the nature of a global
environment. Every test I've seen using a global environment uses it to set up
some expensive external resource, not something that can/should be set up for
each test anew. (Again this is unsurprising, since the environment is a global.)
PiperOrigin-RevId: 424003937
Change-Id: I9e8a825cb8900960dd65b85fe5ffcc0a337e57f3
unifying the behavior between Bazel and CMake
This fixes one of the CI failures on Windows
PiperOrigin-RevId: 417872531
Change-Id: I156989323b7e6d4a4420f4f9691b078829db933d
On KitKat, calling tzset with UTC+nn doesn't initialize all the
timezone state. If the previous timezone was something like
America/Chicago, then changing it to UTC+nn might have no effect.
Setting the timezone to an intermediate value like "UTC" avoids the
problem.
Works around https://github.com/android/ndk/issues/1604.
PiperOrigin-RevId: 413050236
Change-Id: I99b2d3330ae68f1d58cd2ca278d3eaae30bd1e83
Do not attempt to continue running a test suite if it already failed during
`SetUpTestSuite`.
The suite already failed and running the tests might just add noise to the run, or even crash the process unnecessarily.
Fixes#2187
PiperOrigin-RevId: 397770405
gtest: Add a flag to only set up/tear down test environments once when repeating
Currently when running a test multiple times using `--gtest_repeat` the global
test environment(s) are set up and torn down for each iteration of the test.
When checking for flakes in tests that have expensive dependencies that are set
up in the test environment (subprocesses, external dependencies, etc) this can
become expensive.
To support finding flakes in tests that fit into this category, where the setup
phase is expensive but each test case is fast, allow callers to specify via
`--gtest_recreate_environments_when_repeating=false` that the test environments
should only be set up once, for the first iteration, and only torn down once, on
the last iteration. This makes running a test with `--gtest_repeat=1000` a much
faster and more pleasant experience.
PiperOrigin-RevId: 382748942
Make multiple attempts to verify GetThreadCount()
Testing GetThreadCount() is inheritently noisy, as other threads can be started
or destroyed between two calls to GetThreadCount(). This is especially true
under certain analyzer configurations, such as TSAN.
PiperOrigin-RevId: 381951799
Add support to run gtest on Xtensa platform.
This add support to run GTest base test suits on Xtensa
(https://ip.cadence.com/ipportfolio/tensilica-ip) base simulator.
Xtensa only provides libc and some basic operations and does not run an
operating system by default.
PiperOrigin-RevId: 368162205
Print std::u8string, std::u16string, and std::u32string as string literals
Previously, these types were printed as "{ U+123, U+456, U+789 }". However,
printed output in that form is difficult to compare against any literals that
might be defined in code. Instead, just treat these types like std::string
and std::wstring, escaping non-ASCII characters with a hexadecimal escape
sequence.
The tests have also been updated to cover the new functionality: as a bonus,
the tests now also pass with the MSVC toolchain.
Internally, the code has been reorganized to primarily operate in terms of
char32_t, under the assumption that char32_t will always be at least as big
as wchar_t. While that assumption is currently true, perhaps it won't be in
the future...
PiperOrigin-RevId: 364033132
gtest: Output a canned test suite for environment failures in XML/JSON
This surfaces useful information about the environment failure in a structured form.
As we can see from the updated test, previously unsurfaced information is now present.
PiperOrigin-RevId: 362292322
Included the string returned by their `name()` member function with the output of `PrintTo`.
Typical use case:
std::unique_ptr<AbstractProduct> product = FactoryMethod();
// Assert that the product is of type X:
ASSERT_EQ(std::type_index{typeid(*product)},
std::type_index{typeid(ProductX)});
Possible output in case of a test assert failure, now including the names of the compared type indices:
> error: Expected equality of these values:
> std::type_index(typeid(*product))
> Which is: 8-byte object <D0-65 54-8C F6-7F 00-00> ("class ProductY")
> std::type_index(typeid(ProductX))
> Which is: 8-byte object <40-64 54-8C F6-7F 00-00> ("class ProductX")
With help from Krystian Kuzniarek.
Explicitly skip tests after fatal global environment setup errors
Previously the tests were all skipped, but the resulting output claimed all
tests passed.
Before:
```
[----------] Global test environment set-up.
<failure message>
[----------] Global test environment tear-down
[==========] 1 test from 1 test suite ran. (83 ms total)
[ PASSED ] 1 test.
[ FAILED ] 0 tests, listed below:
```
After:
```
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
<failure message>
[----------] 1 test from SomeTest
[ RUN ] SomeTest.DoesFoo
<...>: Skipped
[ SKIPPED ] SomeTest.DoesFoo (0 ms)
[----------] 1 test from SomeTest (0 ms total)
[----------] Global test environment tear-down
[==========] 1 test from 1 test suite ran. (68 ms total)
[ PASSED ] 0 tests.
[ SKIPPED ] 1 test, listed below:
[ SKIPPED ] SomeTest.DoesFoo
[ FAILED ] 0 tests, listed below:
```
PiperOrigin-RevId: 358026389
Print unique_ptr/shared_ptr recursively.
Given that they are smart pointers, it is unlikely that the inner object is
invalid.
PiperOrigin-RevId: 351586888
