dymok93 / ktl Goto Github PK

As a result, operator = (const wchar_t *) and moving from the string may end up in an inconsistent state: capacity ()> SSO_CH_BUFFER_COUNT (i.e. !is_small()), but m_str.get_first (). Buffer == m_buffer

https://en.cppreference.com/w/cpp/algorithm/equal

There may be used assert: https://github.com/DymOK93/KTL/blob/master/include/vector.hpp#L360

• Different types of WorkItems
• DPC
• User-defined pools

Hmm... How about executors?

Let's try porting fmtlib into Kernel!

std::chrono analog with providing access to the kernel clocks

https://github.com/DymOK93/KTL/blob/feature_vector/include/string.hpp#L497
basic_winnt_string::pop_back() calls decrease_size() that doesn't take 3 arguments (second argument of this call is duplicate of the first)

https://github.com/DymOK93/KTL/blob/master/include/string_algorithm_impl.hpp#L47
There must be pointer of type CharT

https://github.com/DymOK93/KTL/blob/feature_extended_readme/runtime/src/exc_engine/member_ptr.cpp

The use of static objects with non-trivial initialization in drivers (Scott Meyers' singleton, for example) leads to BSOD.

__GSHandlerCheck, contrary to the documentation, is added by the compiler to the table of exception handlers and is called even with "/ EHsc". It expects as a parameter a pointer to an object of the _DISPATCHER_CONTEXT structure, the contents of which are different from ktl::crt::exc_engine::dispatcher_constext, which leads to BSOD

https://github.com/DymOK93/KTL/blob/feature_vector/include/string.hpp#L934

Managing threads in the kernel in C++ way

https://en.cppreference.com/w/cpp/algorithm/move_backward

https://github.com/DymOK93/KTL/blob/feature_vector/include/atomic.hpp#L93
https://github.com/DymOK93/KTL/blob/feature_vector/include/atomic.hpp#L104
https://github.com/DymOK93/KTL/blob/feature_vector/include/atomic.hpp#L127
These template specializations must be marked inline to get weak external linkage

https://github.com/DymOK93/KTL/blob/feature_vector/include/string.hpp#L970
If symbol not found traits_type::find() returns nullptr (not a position behind size())

Function pointers are default contructible, but nullptr is not quite what is expected

https://github.com/DymOK93/KTL/blob/feature_vector/include/hash.hpp#L100

https://github.com/DymOK93/KTL/blob/master/runtime/include/utility_impl.hpp#L27 and https://github.com/DymOK93/KTL/blob/master/runtime/include/utility_impl.hpp#28 was messed up

https://github.com/DymOK93/KTL/blob/feature_vector/include/string_view.hpp#L263
find_first_of() must return a position, not a boolean value

Unused structure template (used for type checking by way of a compiler error message )
https://github.com/DymOK93/KTL/blob/feature_vector/modules/lockfree/node_allocator.hpp#L15

https://github.com/DymOK93/KTL/blob/feature_vector/include/string.hpp#L920
There should be used throw_exception_if(pos > size())

https://github.com/DymOK93/KTL/blob/master/runtime/src/object_management.cpp#L19
__cxx_ctors_begin__ and __cxx_ctors_end__ are x64-specific: https://github.com/DymOK93/KTL/blob/master/runtime/include/crt_runtime.h#L12, https://github.com/DymOK93/KTL/blob/master/runtime/include/crt_runtime.h#L25

https://github.com/DymOK93/KTL/blob/feature_vector/include/smart_pointer.hpp#L870
add_lvalue_reference was used instead of add_lvalue_reference_t

If there is only 1 non-trivial object on the stack, destruction takes place normally
Otherwise, with N + 1 objects on the stack, destructors are called only for N

https://github.com/DymOK93/KTL/blob/feature_vector/include/string.hpp#L1060
https://github.com/DymOK93/KTL/blob/feature_vector/include/string.hpp#L1067
RandomAcceccIt instead of the RandomAccessIt

https://github.com/DymOK93/KTL/blob/master/include/hash_table_impl.hpp#L98
There must be exchange(other.mListForFree, nullptr)

https://github.com/DymOK93/KTL/blob/master/include/algorithm.hpp#L372
memcmp() may return any negative or positive values, not only -1 or 1

It is required to explicitly include <new_delete.h> into <memory.hpp>