C++/WinRT configuration macros
This topic describes the C++/WinRT configuration macros. Unless otherwise noted, these rules apply to all of the C++/WinRT configuration macros:
- All files that are linked together to form a single module (
.exeor.dll) must have identical macro settings. That includes static libraries. - All macro settings must be complete before including any C++/WinRT header file.
- You may not change any macro setting after including any C++/WinRT header file.
WINRT_LEAN_AND_MEAN
If defined, disables these rarely-used features (in order to reduce compile times):
- The ability to implement exclusive interfaces outside the component.
- std::hash specializations for interface and runtime class smart pointers.
- Support for directly outputting an hstring or IStringable to a C++ stream, since version 2.0.221101.3.
You're allowed to combine files with different settings for WINRT_LEAN_AND_MEAN.
Files that don't define WINRT_LEAN_AND_MEAN gain access to the rarely-used features.
WINRT_NO_MODULE_LOCK
If defined, disables object counts for the current module. The module never unloads from the process. Defining this macro is customary for executables (which can never unload), or for .dlls that you intend to leave pinned. May not be combined with WINRT_CUSTOM_MODULE_LOCK.
WINRT_CUSTOM_MODULE_LOCK
If defined, allows you to provide your own implementation of winrt::get_module_lock. May not be combined with WINRT_NO_MODULE_LOCK.
Your custom implementation of winrt::get_module_lock must support the following operations:
++winrt::get_module_lock(): Increment the reference count on the module lock.--winrt::get_module_lock(): Decrement the reference count on the module lock.if (winrt::get_module_lock()): Check whether the reference count is nonzero. (Needed if you're building a DLL.)
WINRT_ASSERT, WINRT_VERIFY
These macros allow you to customize assertion handling. WINRT_ASSERT doesn't require the argument to be evaluated. WINRT_VERIFY requires that the argument be evaluated, even in non-debug builds.
If you don't customize these macros, and _DEBUG is defined, then C++/WinRT makes them equivalent to _ASSERTE.
If you don't customize these macros, and _DEBUG is not defined, then C++/WinRT defines WINRT_ASSERT to discard the expression unevaluated, and defines WINRT_VERIFY to discard the expression after evaluating it.
WINRT_NO_MAKE_DETECTION
If defined, disables the default C++/WinRT diagnostic that detects that you mistakenly constructed an implementation class without using winrt::make.
We strongly recommended that you don't define this symbol, because doing so masks a common source of programming errors.
WINRT_NO_SOURCE_LOCATION
If defined, disables the inclusion of source file and line number information (and in debug builds, function information), when originating errors.
This additional information is not used by C++/WinRT, but it is made available to other libraries which wish to interoperate with C++/WinRT exceptions, such as the Windows Implementation Library.
By default, the information is included when compiled in C++20 mode or higher. You may wish to suppress this information to reduce binary size.
WINRT_DIAGNOSTICS
If defined, enables internal statistics to track various operations:
- The number of times each interface was queried.
- The number of times each factory was requested (and whether the factory is agile).
WINRT_NATVIS
If defined, includes helper functions to assist in native debug visualizations in Visual Studio. The code isn't used at runtime; it exists only for debugging.
If you don't customize this macro, then visualization support functions are enabled if _DEBUG is defined. For more details, see Visual Studio native debug visualization (natvis) for C++/WinRT.
You're allowed to combine files with different settings for WINRT_NATVIS.
If any file is compiled with WINRT_NATVIS support, then the resulting module will have native debug visualizations enabled.
WINRT_EXPORT, WINRT_FAST_ABI_SIZE
Don't use these macros.