The original implementation of `UNSCOPED_X` message macros used a
clever hack to make the original implementation simpler: construct
an instance of `ScopedMessage` to manage its lifetime, but store
it in a vector, so its lifetime is not actually scope-based, and
we can manage it through the vector instance.
This hack made it so that the lifetime of the vector that manages
the fake `ScopedMessage`s must be outlived by the vector with the
actual messages. Originally this wasn't a problem, because they both
lived inside the run context instance. However, since then these
vectors became globals and thread-local. When this happened, it
still wasn't a problem; the two globals were declared in the right
order, so they were destroyed in the right order as well.
Then, in f80956a43a, these globals
were turned into magic static globals to improve their behaviour
in MSVC's Debug build mode. This caused their lifetimes to be
runtime-dependent; if a specific test thread added its first scoped
message before it added first unscoped message, the lifetimes
would be correct. If it instead added first unscoped message
before adding first scoped message, then there **might** be
invalid reads during thread destruction.
The fix is simple: do things properly and manage the lifetime of
messages in `UNSCOPED_X` explicitly. Then we don't have to deal
with the destruction of fake `ScopedMessage`s while the thread is
being destroyed, and the lifetime of the two vectors is no longer
tied together.
I also threw them both into a new type, to encapsulate some of the
unscoped message logic.
This avoids calling the global's constructor on threads that will
never interact with them. Calling the constructor can have surprising
overhead, as e.g. MSVC's Debug mode `std::vector` will allocate in
the default constructor.
Closes#3050
Specifically, this commit makes the `-c`/`--section` parameter
strictly ordered and hierarchical, unlike how it behaved before,
which was a huge mess -- see #3038 for details.
Closes#3038
We still keep the error check in the function, but hide it in an
outlined function inside a .cpp file, to promote inlining of the
retrieval part.
In the future, we should explore two things
1) Skipping over the context retrieval here, allowing direct access.
I currently do not see a way to do this while keeping the
"greppability" of mutable vs immutable accesses that is there now,
but it would help a lot when inlining is not enabled.
2) Removing the error check, to make the function trivially inlinable,
and without branches.
**runtime difference**
| --------- | Debug | Release |
|:----------|------:|--------:|
| Slow path | 0.98 | 1.07 |
| Fast path | 1.04 | 1.08 |
We lost bit of performance on the assertion slow path in debug mode,
but together with the previous commit, it comes out at net zero.
For other combinations, we see 5-10% perf improvement across the
two commits.
This builds on the existing work to make assertion thread safe,
by adding an extra synchronization point in the holder of
`ReusableStringStream`'s stream instances, as those are used to
build the messages, and finishing the move of message scope holders
to be thread-local.
All the previous refactoring to make the assertion fast paths
smaller and faster also allows us to implement the fast paths
just with thread-local and atomic variables, without full mutexes.
However, the performance overhead of thread-safe assertions is
still significant for single threaded usage:
| slowdown | Debug | Release |
|-----------|--------:|--------:|
| fast path | 1.04x | 1.43x |
| slow path | 1.16x | 1.22x |
Thus, we don't make the assertions thread-safe by default, and instead
provide a build-time configuration option that the users can set to get
thread-safe assertions.
This commit is functional, but it still needs some follow-up work:
* We do not need full seq_cst increments for the atomic counters,
and using weaker ones can be faster.
* We brute-force updating the reporter-friendly totals from internal
atomic counters by doing it everywhere. We should properly trace
where this is needed instead.
* Message macros (`INFO`, `UNSCOPED_INFO`, `CAPTURE`, etc) are not
made thread safe in this commit, but they can be made thread safe
in the future, by building on top of this work.
* Add more tests, including with thread-sanitizer, and compiled
examples to the repository. Right now, these changes have been
compiled with tsan manually, but these tests are not added to CI.
Closes#2948
This improves the fast path performance for successful assertions
by about 7%, at the cost of potentially keeping around the message
allocation longer.
This means that:
1) Scoped messages are always removed at the end of their scope,
even if the scope ended due to an exception.
2) Scoped messages outlive section end, if that section's scope is
enclosed in their own.
Previously neither of these were true, which has led to a number
of surprising behaviour, where e.g. this:
```cpp
TEST_CASE() {
try {
INFO( "some info" );
throw std::runtime_error( "ex" );
} catch (std::exception const&) {}
REQUIRE( false );
}
```
would print "some info" as the message for the assertion, while this:
```cpp
TEST_CASE() {
INFO("Hello");
SECTION("dummy") {}
REQUIRE(false);
}
```
would not print out "Hello" as the message for the assertion.
This had an underlying reason, in that it was trying to helpfully
keep the messages around in case of unexpected exceptions, so that
code like this:
```cpp
TEST_CASE() {
auto [input, expected] = GENERATE(...);
CAPTURE(input);
auto result = transform(input); // throws
REQUIRE(result == expected);
}
```
would report the value of `input` when `transform` throws. However,
it was surprising in practice and was causing various issues around
handling of messages in other cases.
Closes#1759Closes#2019Closes#2959
`handleNonExpr` is responsible for handling assertions that do not
result in a decomposable expression, e.g. `REQUIRE_THROWS`, or
`REQUIRE_NOTHROW`.
Running benchmark on these two macros specifically, with
`REQUIRE_THROWS([](){ throw 1; }())`, and `REQUIRE_NOTHROW([](){}())`,
we get these speedups:
| | Debug | Release |
|---------|--------|---------|
| THROWS | 3.69x | 2.10x |
| NOTHROW | 1.18x | 1.05x |
Obviously the actual performance improvement is dependent on
how expensive the expression under test is.
This costs us about 1% perf in Debug build and 3% in Release build,
but it is worth it for more precise information during unexpected
exceptions or fatal errors.
Given a simple test case like this
```cpp
TEST_CASE("Hard fail") {
REQUIRE( 1 == 1 );
REQUIRE( 2 == 2 );
throw 1;
REQUIRE( 3 == 3 );
}
```
Catch2 before this change would report the line info from the
`TEST_CASE` macro as the last seen expression before error. With
this change, it will correctly report the line info from the
`REQUIRE(2 == 2)` assertion as the last seen expression before error.
The fast path allows `RunContext` to skip disabling output redirect,
and notifying the reporters, turning `RunContext::notifyAssertionStarted`
into a no-op. This improves the overall performance of assertion handling
significantly, and also prepares ground for future changes around
assertion handling and thread safety.
For simple 10M assertion run, this improves the running time by ~30%
in Debug build and ~40% in Release build.
For backwards-compatibility reasons, the fast path is disabled
by default. However, none of the first party reporters use the
`assertionStarting` event, so all first party reporters are opted-in.
There were only two places where we used the full `AssertionInfo`
instance in `m_lastAssertionInfo`:
1) when reporting unexpected exception from running a test case
2) when reporting fatal error
because in those two places we do not have access to a real
instance of `AssertionInfo`, but we still need to send one to the
reporters. As a bonus, in both of these places we were already
constructing a fake-ish assertion info, by using the last encountered
source location, but dummying out the other information.
Instead, we only keep track of the last encountered source location,
and construct the dummy `AssertionInfo` on-demand.
This finishes the set of refactoring around `m_lastAssertionInfo`
in `RunContext` and improves the performance of running assertions
by ~5% in both Debug and Release mode.
--------------
Note that this change also causes small difference in output. It
could be avoided by having an invalidation flag and tracking where
the information would be invalidated before, but the difference
includes more precise line location for unexpected errors (both
exceptions and fatals), so I prefer the new output.
Because `RunContext::populateReaction` no longer implicitly depends
on the value of `RunContext::m_lastAssertionInfo`, we don't have to
delay the reset until `RunContext::reportExpr` is finished.
`RunContext::resetAssertionInfo` overwrites everything in
`m_lastAssertionInfo` except for the lineInfo, so if the reset
is called at the end of a copying function, we can save work
by copying only the lineInfo.
This rework changes two important things
1) the output redirect is deactivated while control is given to the reporters.
This means that combining reporters that write to stdout with capturing
reporters, e.g. `./tests -s -r console -r junit::out=junit.xml`, no
longer leads to the capturing reporter seeing all the output from
the other reporter captured.
Trying this with the `SelfTest` binary would previously lead to JUnit
spending **hours** trying to escape all of ConsoleReporter's output and
write it to the output file. I actually ended up killing the process
after 3 hours, during which the JUnit reporter wrote something like 50 MBs
of output to a file.
2) The redirect object's lifetime is tied to the `RunContext`, instead
of being constructed for every partial test case run separately.
This has no effect on the basic StreamRedirect, but improves the FileRedirect
significantly. Previously, running many tests in single process with this
redirect (e.g. running `SelfTest -r junit`) would cause later tests to
always fail before starting, due to exceeding the limit of temporary files.
For the current `SelfTest` binary, the old implementation would lead to
**295** test failures from not being able to initiate the redirect. The
new implementation completely eliminates them.
----
There is one downside to the new implementation of FileRedirect, specific
to Linux. Running the `SelfTest` binary on Linux causes 3-4 tests to have
no captured stdout/stderr, even though the tests were supposed to be
writing there (there was no output to the actual stdout/stderr either,
the output was just completely lost).
Since this never happen for smaller test case sets, nor does it reproduce
on other platforms, this implementation is still strictly better than
the old one, and thus it can get reasonably merged.
This PR introduces a new `TEST_CASE` macro called `TEST_CASE_PERSISTENT_FIXTURE`. `TEST_CASE_PERSISTENT_FIXTURE` offers the same functionality as `TEST_CASE_METHOD` except for one difference. The object on which the test method is invoked is only created once for all invocations of the test case. The object is created just after the `testCaseStarting` event is broadcast and the object is destroyed just before the `testCaseEnding` event is broadcast.
The main motivation for this new functionality is to allow `TEST_CASE`s to do expensive setup and teardown once per `TEST_CASE`, without having to resort to abusing event listeners or static function variables with manual initialization.
Implements #1602
---------
Co-authored-by: Martin Hořeňovský <martin.horenovsky@gmail.com>
Closes#1210
When a signal is caught, the destructors of Sections will not be called.
Thus, we must call `sectionEndedEarly` manually for those Sections.
Example test case:
```
TEST_CASE("broken") {
SECTION("section") {
/// Use illegal cpu instruction
__asm__ __volatile__("ud2" : : : "memory");
}
}
```
Classes will automatically inherit the virtual-ness of their base
class destructors. If the base class already has a virtual
destructor and the derived class needs default destructor semantics
then the derived class can omit defining the destructor in favor of
the compiler automatically defining it.
This has an additional benefit of reenabling move semantics. The
presence of a user-specified destructor automatically disables move
operations.
* AssertionEnd does not reset the assertion info yet. That is done after populateReaction. And reset assertion info would also reset the result disposition to normal, so that any uncaught exception would be reported as failure
* Approving test output changes due to added unit tests
* Unit tests to throw std::runtime_error instead of std::exception
* Add a unit test to test incomplete assertion handler
---------
Co-authored-by: Ross <ross.tang@gfo-x.com>
* Split out BenchmarkInfo and BenchmarkStats to their own header
* Outline BenchmarkStats<> declaration to separate header
* Split out TestRunInfo into its own header
These changes let us remove the large `interfaces_reporter.hpp`
include from `benchmark.hpp`, and replace it with
`interfaces_capture.hpp` in `run_context.hpp`.
I also cleaned out `interfaces_repoter.hpp` from reporter headers
that depend on `reporter_common_base.hpp`. This will not change
anything in the actual inclusion set, but makes it logically
more consistent.
`NameAndLocationRef` is pretty large type, so even in release build,
it is unlikely to be passed in registers. In addition to the fact
that some platforms currently do not allow passing even small types
in register (Windows ABI!!), it is better to pass it as a ref,
effectively passing around a pointer.
This eliminates 1945 (432709 -> 430764) allocations from running
`./tests/SelfTest -o /dev/null`. In general terms, this saves
an allocation every time an unvisited `SECTION` is passed, which
means that the saved allocations are quadratic in number of sibling
(same level) `SECTION`s in a test case.
Now we delay allocating owning `NameAndLocation` instances until
we construct a new tracker (because a tracker's lifetime can be
significantly different from the underlying tracked-thing's name).
This saves 4239 allocations (436948 -> 432709) when running
`./tests/SelfTest -o /dev/null`, at some cost to code clarity
due to introducing a new ref type, `NameAndLocationRef`.
When running `./tests/SelfTest -o /dev/null`, this saves 1272
allocations (437439 -> 437167). In general, this saves multiple
allocations per end of an entered `SECTION`, if the section name
was too long for SSO, because `RunContext::sectionEnded` can then
move the section's name further down the callstack.
