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
In case of 2 instances of SourceLineInfo constructed in the same
file, they will have the same `file` pointer (even at O0). Thus, we
can check if they are equal before calling potentially pointless
`strcmp`.
In theory the copy is cheap (couple of pointers change), but tests
are usually compiled in Debug mode/with minimal optimizations, which
means that most users will still have to pay the cost for those
function calls.
Because the macro name is compile-time constant, we do not have to
worry about lifetimes and will avoid allocation in case of missing
SSO or long macro name.
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
- seems the #ifdef was necessary after all, because of the difference in the way the cpp files are included in the full project vs the single include
- in the OC project I moved the #include of catch_tostring.cpp first. That solves the project for now, but is a brittle solution
Happening when using clang and templated operators, clang cannot decide
between the operator provided by ReusableStringStream and the one provided
by the value value as both are templates. This is easily solved by calling
the operator<< through the member syntax.
Fixes#1285
This was removed in 64be2ad, to fix OS X approval tests. At the time
I couldn't investigate because I didn't have access to OS X, but this
fixed it (and since we don't have MinGW in CI, the breakage went
unnoticed).
As it turns out, piece-wise compilation of the Compact
reporter had broken OS X detection for a long time, and fixing it
was what broke the approvals. After the approval scripts were
changed to compensate, this change passes approval tests and fixes
Until now, the stack size for POSIX signal handling was determined by
the implementation defined limit `STKSZ`, which in some cases turned out
to be insufficient, leading to stack overflow inside the signal handler.
The new size, which was determined experimentally, is the larger of 32kb
or `MINSTKSZ`.
Fixes#1225
This should align more closely with the intended semantics, where
types without `StringMaker` specialization or `operator<<` overload
are passed down to the user defined fallback stringifier.
Related to #1024
This support is based on overriden `std::exception::what` method, so
if an exception does not do so meaningfully, the message is still
pointless.
This is only used as a fallback, both `StringMaker` specialization and
`operator<<` overload have priority..
Android apparently does not support `std::to_string`, so we add a
small polyfill over it. Right now only the ULP matcher uses it,
but we have had plans to use it in `StringMaker<int>` and friends,
as it performs a lot better than `std::stringstream` based
stringification on MSVC.
See #1280 for more details
While the comment format was valid C++, it breaks our tooling badly.
I opened up a github issue for our tooling, because unexpected
formatting of a block comment should not silently generate invalid
single header file, see #1269.
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
std::isalnum expects an int in the range of unsigned char or -1 (EOF),
otherwise it exhibits undefined behavior. Casting from char to unsigned
char avoids this.
MSVC warns about this when compiling with /analyze.
Catch2's documentation promises that listeners are called _before_
reporters, but because of the previous implementation, they were
called _after_ reporters. This commit fixes that.
Closes#1234
The fix leaves an open question: should we keep treating refs
to static array of chars as strings, or should we instead
use `strnlen` to check if it is null-terminated within the buffer
Fixes#1238
Specific platforms (e.g. TDM-GCC) can have terrible timer resolution,
and our checking code will then loop for an inordinate amount of time.
This change will make it so that the calibration gives up after 3
seconds and just uses the already measured values.
This leaves one open question, how to signal that the resolution
is terrible and benchmarking should not happen?
Fixes#1237
There are still some holes, e.g. we leave surrogate pairs be
even though they are not a part of valid UTF-8, but this might
be for the better -- WTF-8 does support surrogate pairs inside
text.
Closes#1207
DJGPP cross compiler is targeting DOS which does not support POSIX
signals. Probably for the same reason (targeting DOS) this compiler
does not support wide characters.
This allows reducing the amount of friends needed for its interface
and some extra tricks later.
The bad part is that the pointer can become invalidated via
calls to other StringRef's public methods, but c'est la vie.
To prevent bugs with stitching system headers inside Catch,
the proxy header is responsible for guarding against inclusion
on Linux, rather than the includers.
Might be related to #1197
