8475061136
thanks to Nico Werner, who did most of the porting work
12 KiB
12 KiB
Roadmap
This library is free, and will stay free but needs your support to sustain its development. There are lots of new features and maintenance to do. If you work for a company using doctest or have the means to do so, please consider financial support.
Planned features for future releases - order changes constantly... Also look through the issues.
For 2.5:
- https://github.com/onqtam/doctest/issues/208
- reporters:
- ability to redirect the stdout/stderr from test cases and capture it - like Catch does it
- xUnit/TeamCity reporter
- compact reporter
- matchers - should investigate what they are - look at google test/mock and Catch (also predicates and boost test)
- header with extensions
- demangling with the use of the cxxabi header
- stringification of types from std, also enums with the help of traits as discussed in #121
- esoteric reporters
- convolution support for the assertion macros (with a predicate)
- Value-Parameterized test cases
- generators? - look at Catch - and investigate what they are (also SUBCASEs can be while() loops instead of if() statements! that might be useful...)
- look at property based testing
- proper conan package - https://github.com/onqtam/doctest/issues/103
- IDE integration
- https://blogs.msdn.microsoft.com/vcblog/2017/05/10/unit-testing-and-the-future-announcing-the-test-adapter-for-google-test/
- https://www.reddit.com/r/cpp/comments/65c0f1/run_cpp_unit_tests_from_xcode_and_visual_studio/
- https://github.com/k-brac/CUTI
- https://github.com/csoltenborn/GoogleTestAdapter
- MSTest
- XCode - https://github.com/catchorg/Catch2/pull/454
- CLion
For 2.6:
- log levels - like in boost test
- running tests a few times
- test execution in separate processes -
fork()
for UNIX and this for Windows - killing a test that exceeds a time limit (will perhaps require threading or processes)
- symbolizer - for a stack trace - when an assertion fails - and it's in a user function with some deep callstack away from the current test case - how to know the exact code path that lead to the failing assert
- ability to make the framework not capture unexpected exceptions - as requested here
- add Approx ability to compare with absolute epsilon - Catch PR
- ability to customize the colors in the console output (may also use styles - based on this or this)
- implement breaking into the debugger under linux - see here and here
- better testing of the library
- unit test the String class
- should unit test internals - currently even if a bug is caught by different output it's very difficult to track the reason
- should test stuff that should not compile
- should test crash handling
- should test more config options
- don't cheat for maxing out code coverage (see coverage_maxout.cpp)
- should test C++11 stuff - perhaps inspect the CMAKE_CXX_FLAGS for -std=c++11 on the CI and add more targets/tests
- test tricky stuff like expressions with commas in asserts
For 3.0:
- use modules - use
std::string
and whatever else comes from the standard - no more hand rolled traits and classes - minimize the use of the preprocessor
- remove backwards-compatible macros for the fast asserts
Things that are being considered but not part of the roadmap yet:
- add ability to print a diff for strings/values instead of the whole 2 versions (gtest does that)
- add LIKELY & friends for the conditions of asserts - look at BOOST_LIKELY, ppk_assert, foonathan/debug_assert, etc
- ability for users to register their own command line options and access them later on
- fix this: https://github.com/catchorg/Catch2/issues/1292
- FakeIt mocking integration - like catch (also checkout this)
- look into https://github.com/cpp-testing/GUnit - https://www.youtube.com/watch?v=NVrZjT5lW5o
- consider the following 2 properties for the MSVC static code analyzer: EnableCppCoreCheck, EnableExperimentalCppCoreCheck
- rpm package? like this: https://github.com/vietjtnguyen/argagg/blob/master/packaging/rpm/argagg.spec
- get the current test case/section path - https://github.com/catchorg/Catch2/issues/522
- when no assertion is encountered in a test case it should fail - and should also add a SUCCEED() call
- failure reporting should print out previous SECTIONs for data-driven testing - as requested here
Bitwise()
class that has overloaded operators for comparison - to be used to check objects bitwise against each other- detect floating point exceptions
- checkpoint/passpoint - like in boost test (also make all assert/subcase/logging macros to act as passpoints and print the last one on crashes or exceptions)
- queries for the current test case - name (and probably decorators)
- support for LibIdentify
- add CHECKED_IF & friends: https://github.com/catchorg/Catch2/issues/1278
- support for running tests in parallel in multiple threads
- death tests - as in google test
- config options
- test case name uniqueness - reject the ones with identical names
- command line options
- ability to specify ASC/DESC for the order option
- global timeout option (per test or per entire session?)
- command line error handling/reporting
- option to not print context info when the --success option is used
- ability for the user to extend the command line - as requested here
- option to list files in which there are test cases who match the current filters
- option for filters to switch from "match any" to "match all" mode
- option to list test suites and test cases in a tree view
- add a "wait key" option (before and after tests) - as requested here
- decorators for test cases and test suites- like in boost test
- depends_on
- precondition
- fixture
- label (tag) - with the ability to have multiple labels (tags) for a test case and also the ability to list them
- run X times (should also multiply with (or just override) the global test run times)
- throw an exception when incompatible decorators are given in the same list of decorators - like may_fail and should_fail
- setup / teardown support
- global setup / teardown - can be currently achieved by providing a custom main function
- per test suite (block? only? and not all blocks of the same test suite?)
- as decorators
- see how it's done in boost test - with the fixture decorator
- perhaps for fixtures in addition to the constructor / destructor - since throwing in the destructor might terminate the program
- or just ignore all of this this - it would require globals or classes and inheritance - and we already have subcases
- doctest in a GUI environment? with no console? APIs for attaching a console? querying if there is one? investigate...
- runtime performance
- look at this: https://github.com/catchorg/Catch2/issues/1086
- startup - the set holding all registered tests should use a specialized allocator to minimize program startup time
- optimize the mutex lock:
- ability to provide a temp folder that is cleared between each test case
- make the _MESSAGE assert macros work with variadic arguments - and maybe write the ones for binary/unary asserts as well
- move from operator "<<" to "<=" for capturing the left operand when decomposing binary expressions with templates
- think about silencing warnings about unused variables when DOCTEST_CONFIG_DISABLE is used - see commit 6b61e8aa3818c5ea100cedc1bb48a60ea10df6e8 or issue #61
- also this:
(void)(true ? (void)0 : ((void)(expression)))
- also this:
- think about optionally using
<typeinfo>
and libcxxabi for demangling so users don't have to useTYPE_TO_STRING()
- handle more complex expressions -
CHECK(foo() == 1 || bar() == 2);
- add noreturn to MessageBuilder::react() - and actually make a separate function (react2) for the FAIL() case
- think about using a string view of some sorts
- benchmark against google test and boost test
Things that are very unlikely to enter the roadmap:
- rethink static code analysis suppressions - users shouldn't have to use the same flags for code which uses doctest macros/types
- move the "react()" part (the one that throws for REQUIRE asserts - or for when "abort-after=" is reached) to a function call in the while() part of the asserts
- stop using underscores for the beginning of identifiers - the anonymous variables - against the standard...
- templated fixture test cases
- test with missed warning flags for GCC
- utf8 / unicode ???
- handle
wchar
strings??? - hierarchical test suites - using a stack for the pushed ones
- ability to specify the width of the terminal in terms of characters (for example 60 - less than 80 - the default)
- ability to re-run only newly compiled tests based on time stamps using
__DATE__
and__TIME__
- stored in some file - add underscores to all preprocessor identifiers not intended for use by the user
- put everything from the
detail
namespace also in a nested anonymous namespace to make them with internal linkage - ability to put everything from doctest into an anonymous namespace - to allow the use of multiple different versions of doctest within the same binary (executable/dll) - like the stb libraries can