Given that in the 2 or so years that matchers are thing nobody complained, it seems that people do not actually write this sort of code, and the possibility will be removed in v3. However, to avoid correctness bugs, we will have to support this weird code in v2.
4.4 KiB
Deprecations and incoming changes
This page documents current deprecations and upcoming planned changes inside Catch2. The difference between these is that a deprecated feature will be removed, while a planned change to a feature means that the feature will behave differently, but will still be present. Obviously, either of these is a breaking change, and thus will not happen until at least the next major release.
Deprecations
--list-*
return values
The return codes of the --list-*
family of command line arguments
will no longer be equal to the number of tests/tags/etc found, instead
it will be 0 for success and non-zero for failure.
--list-test-names-only
--list-test-names-only
command line argument will be removed.
ANON_TEST_CASE
ANON_TEST_CASE
is scheduled for removal, as it can be fully replaced
by a TEST_CASE
with no arguments.
Secondary description amongst tags
Currently, the tags part of TEST_CASE
(and others) macro can also
contain text that is not part of tags. This text is then separated into
a "description" of the test case, but the description is then never used
apart from writing it out for --list-tests -v high
.
Because it isn't actually used nor documented, and brings complications to Catch2's internals, description support will be removed.
SourceLineInfo::empty()
There should be no reason to ever have an empty SourceLineInfo
, so the
method will be removed.
Composing lvalues of already composed matchers
Because a significant bug in this use case has persisted for 2+ years without a bug report, and to simplify the implementation, code that composes lvalues of composed matchers will not compile. That is, this code will no longer work:
auto m1 = Contains("string");
auto m2 = Contains("random");
auto composed1 = m1 || m2;
auto m3 = Contains("different");
auto composed2 = composed1 || m3;
REQUIRE_THAT(foo(), !composed1);
REQUIRE_THAT(foo(), composed2);
Instead you will have to write this:
auto m1 = Contains("string");
auto m2 = Contains("random");
auto m3 = Contains("different");
REQUIRE_THAT(foo(), !(m1 || m2));
REQUIRE_THAT(foo(), m1 || m2 || m3);
Planned changes
Reporter verbosities
The current implementation of verbosities, where the reporter is checked up-front whether it supports the requested verbosity, is fundamentally misguided and will be changed. The new implementation will no longer check whether the specified reporter supports the requested verbosity, instead it will be up to the reporters to deal with verbosities as they see fit (with an expectation that unsupported verbosities will be, at most, warnings, but not errors).
Output format of --list-*
command line parameters
The various list operations will be piped through reporters. This means that e.g. XML reporter will write the output as machine-parseable XML, while the Console reporter will keep the current, human-oriented output.
CHECKED_IF
and CHECKED_ELSE
To make the CHECKED_IF
and CHECKED_ELSE
macros more useful, they will
be marked as "OK to fail" (Catch::ResultDisposition::SuppressFail
flag
will be added), which means that their failure will not fail the test,
making the else
actually useful.
Change semantics of [.]
and tag exclusion
Currently, given these 2 tests
TEST_CASE("A", "[.][foo]") {}
TEST_CASE("B", "[.][bar]") {}
specifying [foo]
as the testspec will run test "A" and specifying
~[foo]
will run test "B", even though it is hidden. Also, specifying
~[baz]
will run both tests. This behaviour is often surprising and will
be changed so that hidden tests are included in a run only if they
positively match a testspec.
Console Colour API
The API for Catch2's console colour will be changed to take an extra argument, the stream to which the colour code should be applied.
Type erasure in the PredicateMatcher
Currently, the PredicateMatcher
uses std::function
for type erasure,
so that type of the matcher is always PredicateMatcher<T>
, regardless
of the type of the predicate. Because of the high compilation overhead
of std::function
, and the fact that the type erasure is used only rarely,
PredicateMatcher
will no longer be type erased in the future. Instead,
the predicate type will be made part of the PredicateMatcher's type.