Question

I’m working already a good time with the .net framework. In this time there were some situations, I used P/Invoke for doing things that I couldn’t do with managed code.

However I never knew exactly what the real disadvantages of using it were. That was also the reason why I tried not to use it as far as possible.

If I google, then I find various posts about it, some draw a catastrophic picture and recommend never to use it. What are the major drawbacks and problems of an app that uses one ore more P/Invoke-calls and when do they apply. (not of a performance perspective, more in the manner “could not be executed from a network share”)?

Answer 1

1. Marshalling between managed/unmanaged types has an additional overhead
2. Doesn’t work in medium trust
3. Not quite intuitive and could lead to subtle bugs like leaking handles, corrupting memory, …
4. It could take some time before getting it right when looking at an exported C function
5. Problems when migrating from x86 to x64
6. When something goes wrong you can’t simply step into the unmanaged code to debug and understand why you are getting exceptions. You can’t even open it in Reflector 🙂
7. Restricts cross platform interoperability (ie: you can’t run under Linux if you rely on a Windows-only library).

Conclusion: use only if there’s no managed alternative or in performance critical applications where only unmanaged libraries exist to provide the required speed.