#pragma managed(push, on) and #pragma managed(push, off) in C++/CLI?

Question

What is the difference between #pragma managed(push, off) and #pragma managed(push, on)?

When we need to use #pragma managed(push, on) instead of #pragma managed(push, off)?

Answer 1

It's obvious - "off" turns managed code off, "on" - turns it on.
We use #pragma managed(push, on) when we need to save state and turn it on.
In common case, in header file, we don't know if managed code enabled, so we enable it:

#pragma once
#pragma managed(push, on)
public ref class Foo
{
    ...
};
#pragma managed(pop)