本文探讨了C++ Interop相较于标准PInvoke的优势与不足。优势包括：无需使用DllImport语句，通过C/C++头文件引用非托管函数；性能略高，因为开发者可以控制数据转换与内存分配等操作；能进一步优化被多个非托管函数使用的数据的传递过程。不足之处在于，当非托管函数需要处理复杂数据类型时，开发者需自行完成双向的数据转换工作；此外，除非将C++ Interop调用包装在托管包装器中，否则数据转换代码通常会与业务逻辑混杂在一起。
The following are the advantages of C++ Interop:
• You don’t need to use DllImport statements. You reference the unmanaged functions
using the standard C/C++ header files.
• C++ Interop is slightly faster than standard PInvoke. This is because you are controlling
the marshaling of data and are able to see when memory is being allocated, data is
being converted, and so on. You have unlimited opportunities to optimize the marshaling
code since it is completely under your control.
• You can further optimize marshaling of data that may be used by multiple unmanaged
functions. By marshaling the data once, you may be able to pass the same data to each
function, eliminating the marshaling normally done for each function call.
And here are the disadvantages of C++ Interop:
• Did I already say that you have to do the marshaling? If the function requires anything
other than blittable data types, you are responsible for all marshaling of data in both
directions. The additional code required on your part might be significant compared to
letting PInvoke do the work for you.
• Unless you wrap the C++ Interop calls in a managed wrapper, the marshaling code is
generally inline with your business logic. In contrast, PInvoke hides the complexity of
marshaling under the covers and allows you to concentrate on solving your business
problems.
扫一扫
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
