It checks Knuth's _close enough with tolerance_ relationship, that
is `|lhs - rhs| <= epsilon * max(|lhs|, |rhs|)`, rather then the
_very close with tolerance_ relationship that can be written down as
`|lhs - rhs| <= epsilon * min(|lhs|, |rhs|)`.
This is because it is the more common model around the internet, and
as such is likely to be less surprising to the users. In the future
we might want to provide the other model as well.
Closes#1746
This means that if you nest multiple random generators inside one
test case, they will not return the same sequence of numbers.
Idea taken from #1736 by Amit Herman.
Closes#1736Closes#1734
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.
Wrong nesting of namespaces resulted in the `Catch` namespace
being ambigous between `::Catch` and `::{anon}::Catch` namespaces.
This should fix it.
Closes#1761
It was used in checking that types in TEMPLATE_TEST_CASE and friends
were unique, but this was removed for v2.8.0 (#1628). Since there
are no further uses of this trait, the simplest thing to do is to
just remove it.
Fixes#1757
The leading/trailing whitespace is problematic because of e.g.
`WHEN` macro having preceeding whitespace for alignment, and it is
generally messy.
Credits to Phil who did lot of the original work.
Closes#1708
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.
The old code was a left-over from the times when the
`capturedExpression` member was a `const char*`, which could always
be indexed. With the change to use `StringRef`, blindly indexing 0th
element is invalid, as it is not indexable part of a StringRef.
The parameter given to `convert` may not be copyable therefore it has to be
captured by const reference. For example an `std::tuple` that contains a
non-copyable type is itself non-copyable.
The NonDefaultConstructible test-case was reduced by one example type
because it did not add any value.
`print` version of the logging functions supports `printf`-like
formatting, which we do not use and given our current debug print
internals, will never use. This should be slightly more efficient
and expresses the intent better.
This PR ultimately does 3 things
* Separately tracks matched tests per each filter part (that is, a set of filters separated by an OR (`,`)), which allows Catch2 to report each of the alternative filters that don't match any tests.
* Fixes `-w NoTests` to return non-zero in the process
* Adds tests for `-w NoTests`.
Noticed that the code was originally concatenating strings just to
then append the result to another string. Now it does not create
temporaries and also preallocates the string buffer.
* Fix non-default-constructible type lists used in TEMPLATE_LIST_TEST_CASE
std::tuple is not default constructible when the first type is not
default-constuctible. Therefore it can not be instantiated.
to circumvent this, we have to use std::declval in the unevaluate decltype
context.
This keeps it out of the main include path when benchmarking is
enabled, somewhat reducing the compilation-time penalty.
Also moved some other functions into the .cpp file, especially
helpers that could be given internal linkage, and concretized some
iterator-templated code that only ever used
`std::vector<double>::iterator`.
This allows us to move <stdexcept> out of the common path, and replace
it with just <exception>. The difference between these two headers is
~13k lines after preprocessing on libstdc++ (16k vs 3k) and ~17k lines
for MS's STL(33k vs 16k).
Note that this is only beneficial if no other stdlib header we use
includes <stdexcept>. AFAIK this is true for the newest MS's STL,
but I have no idea of the applicability for libstdc++ and libc++.
* 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 fixes an issue where a self-assignment of a StringRef copy would point into internally (and now dangling) data.
(now self-assignment check is no longer needed)
Previously we had them to avoid including <algorithm> in the vector
matchers, but
* we included it anyway, even though we did not use it
* we use <algorithm> anyways in the generators
Eventually this needs to be fixed in the textflow project by Phil,
but he has not done so in the half a year this bug has been known
to be there, so...
Closes#1470Closes#1455
* Deduce map return type implicitly
Giving the first template argument to map generator function to deduce
return type is now optional even if the return type is different from
the type generated by mapped generator.
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 generator collects values from the underlying generator until it
has a specified amount of them, and then returns them in one "chunk".
In case the underlying generator does not have enough elements for
a specific chunk, the left-over elements are discarded.
Closes#1538
7f229b4f caused the output file to get opened twice, while
some types of files (e.g. named pipes) can be only opened once.
After this change Session::applyCommandLine opens the output file
only when there is an error to print.
Previously, for a TEMPLATE_PRODUCT_TEST_CASE("Test" ..., T, (P1, P2)),
the generated test case names were
Test - 0
Test - 1
With this commit, the correct typename is used:
Test - T<P1>
Test - T<P2>
-----------
MSVC needs another indirection to evaluate INTERNAL_CATCH_STRINGIZE
and also inserts a space before theINTERNAL_CATCH_STRINGIZE_WITHOUT_PARENS
parameter, which we can get rid of by pointer arithmetic.
The REQUIRE_THROWS and CATCH_REQUIRE_THROWS macros have
a subtle inconsistency in their implementation which can
cause a warning if [-Wunused-value] is used.
This commit changes CATCH_REQUIRE_THROWS so it has the
same implementation as REQUIRE_THROWS
It looks like REQUIRE_THROWS was change in commit
fae0fa4ec but not CATCH_REQUIRE_THROWS.
Similar changes for CATCH_CHECK_THROWS
The clock estimator has a potential division by zero.
Using `iteration + 1` seems also more logical to me for
an average.
Found with coverity in a downstream project.
The ostream passed as reference to `hexEscapeChar` is manipulated
and its original state not restored. This fixes it.
Seen via coverity in a downstream project.
