Because new glibc has changed `MINSIGSTKSZ` to be a syscall instead
of being constant, the signal posix handling needed changes, as it
used the value in constexpr context, for deciding size of an array.
It would be simple to fix it by having the handler determine the
signal handling stack size and allocate the memory every time the
handler is being installed, but that would add another allocation
and a syscall every time a test case is entered.
Instead, I split apart the idea of preparing fatal error handlers,
and engaging them, so that the memory can be allocated only once
and still be guarded by RAII.
Also turns out that Catch2's use of `MINSIGSTKSZ` was wrong, and
we should've been using `SIGSTKSZ` the whole time, which we use now.
Closes#2178
The problem was that under specific circumstances, namely that none
of their children progressed, `GeneratorTracker` will not progress.
This was changed recently, to allow for code like this, where a
`SECTION` follows a `GENERATE` at the same level:
```cpp
SECTION("A") {}
auto a = GENERATE(1, 2);
SECTION("B") {}
```
However, this interacted badly with `SECTION` filters (`-c foo`),
as they could deactivate all `SECTION`s below a generator, and thus
stop it from progressing forever. This commit makes GeneratorTracker
check whether there are any filters active, and if they are, it checks
whether its section-children can ever run.
Fixes#2025
This means that code such as
```cpp
TEST_CASE() {
SECTION("first") { SUCCEED(); }
auto _ = GENERATE(1, 2);
SECTION("second") { SUCCEED(); }
}
```
will run and report 3 assertions, 1 from section "first" and 2
from section "second". This also applies for greater and potentially
more confusing nesting, but fundamentally it is up to the user to
avoid overly complex and confusing nestings, just as with `SECTION`s.
The old behaviour of `GENERATE` as first thing in a `TEST_CASE`,
`GENERATE` not followed by a `SECTION`, etc etc should be unchanged.
Closes#1938
* Successive executions of the same `GENERATE` macro (e.g. because
of a for loop) no longer lead to multiple nested generators.
* The same line can now contain multiple `GENERATE` macros without
issues.
Fixes#1913
In the future, we will also want to introduce our own
`uniform_int_distribution` and `uniform_real_distribution` to get
repeatable test runs across different platforms.
This way it is explicit when there is a `StringRef` -> `std::string`
conversion and makes it easier to look for allocations that could
be avoided.
Doing this has already removed one allocation per registered test
case, as there was a completely pointless `StringRef` -> `std::string`
conversion when parsing tags of a test case.
* Units from <ratio> are no longer redeclared in our own namespace
* The default clock is `steady_clock`, not `high_resolution_clock`,
because, as HH says "high_resolution_clock is useless. If you want
measure the passing of time, use steady_clock. If you want user
friendly time, use system_clock".
* Benchmarking support is opt-in, not opt-out, to avoid the large
(~10%) compile time penalty.
* Benchmarking-related options in CLI are always present, to decrease
the amount of code that is only compiled conditionally and making
the whole shebang more maintainble.
Changes done to Nonius:
* Moved things into "Catch::Benchmark" namespace
* Benchmarks were integrated with `TEST_CASE`/`SECTION`/`GENERATE` macros
* Removed Nonius's parameters for benchmarks, Generators should be used instead
* Added relevant methods to the reporter interface (default-implemented, to avoid
breaking existing 3rd party reporters)
* Async processing is guarded with `_REENTRANT` macro for GCC/Clang, used by default
on MSVC
* Added a macro `CATCH_CONFIG_DISABLE_BENCHMARKING` that removes all traces of
benchmarking from Catch
This adds UNSCOPED_INFO macro, creating a log message that is stored
until the end of next assertion or the end of test case, whichever comes
first. These messages are not scoped locally, unlike messages created by
INFO macro.
This means
* Adding new configuration toggle `CATCH_CONFIG_DISABLE_EXCEPTIONS`
and a best-guess configuration auto-checking for it.
* Adding new set of internal macros, `CATCH_TRY`, `CATCH_CATCH_ALL`
and `CATCH_CATCH_ANON` that can be used in place of regular `try`,
`catch(...)` and `catch(T const&)` respectively, while disappearing
when `CATCH_CONFIG_DISABLE_EXCEPTIONS` is enabled.
* Replacing all uses of `throw` with calls to `Catch::throw_exception`
customization point.
* Providing a default implementation for the above customization point
when `CATCH_CONFIG_DISABLE_EXCEPTIONS` is set.
* Letting users override this implementation with their own.
* Some minor changes and ifdefs all around to support the above
The support is to be considered experimental, that is, the interfaces,
the first party generators and helper functions can change or be removed
at any point in time.
Related to #850
By opting the JUnit and XML reporters into it, we no longer run
into problem where they underreport the results without `-s` flag.
Related to #1264, #1267, #1310
Unlike the relatively non-invasive old way of capturing stdout/stderr,
this new way is also able to capture output from C's stdlib functions
such as `printf`. This is done by redirecting stdout and stderr file
descriptors to a file, and then reading this file back.
This approach has two sizeable drawbacks:
1) Performance, obviously. Previously an installed capture made the
program run faster (as long as it was then discarded), because a call
to `std::cout` did not result in text output to the console. This new
capture method in fact forces disk IO. While it is likely that any
modern OS will keep this file in memory-cache and might never actually
issue the IO to the backing storage, it is still a possibility and
calls to the file system are not free.
2) Nonportability. While POSIX is usually assumed portable, and this
implementation relies only on a very common parts of it, it is no
longer standard C++ (or just plain C) and thus might not be available
on some obscure platforms. Different C libs might also implement the
relevant functions in a less-than-useful ways (e.g. MS's `tmpfile`
generates a temp file inside system folder, so it will not work
without elevated privileges and thus is useless).
These two drawbacks mean that, at least for now, the new capture is
opt-in. To opt-in, `CATCH_CONFIG_EXPERIMENTAL_REDIRECT` needs to be
defined in the implementation file.
Closes#1243
ReusableStringStream holds a std::ostringstream internally, but only exposes the ostream interface.
It caches a pool of ostringstreams in a vector which is currently global, but will be made thread-local.
Altogether this should enable both runtime and compile-time benefits. although more work is needed to realise the compile time opportunities.
An empty default implementation is provided to keep backward compatibility.
Called when signal or Structured Exception is encountered.
Related to #1005
Swept:
`-Wpadded` in some places (where it caused extra size, instead of just
saying "hey, we padded struct at the end to align, just as standard says")
`-Wweak-vtables` everywhere (Clang)
`-Wexit-time-destructors` everywhere (Clang)
`-Wmissing-noreturn` everywhere (Clang)
The last three are enabled for Clang compilation going forward.
Also enabled `-Wunreachable-code` for Clang and GCC
