In .NET the Common Language Runtime (CLR) virtual machine manages the program execution. Code that is written in a .NET language is hence called Managed Code, while code like C and C++ that compiles to native programs that doesn't require CLR is referred to as Unmanaged Code.
It turns out that many types can not be shared straight up between managed code and unmanaged code. To do this the types needs to be converted. This conversion process is called marshaling.
I got some real hands on experience with marshaling the last week when I worked on moving and re-writing C++ code to C#. One of the structs (that included an array of another type of struct, that in turn included another struct...) in the C++ code is written to disk with exact requirements on the number of bytes the data is allowed to occupy.
My first naive approach was to simply replicate the structs in C# and, using a BinaryFormatter and a MemoryStream, convert them to a byte array. This approach worked very well in converting back and forth between objects and binary data. Unfortunately for me the size of the byte array became about twice as large as the allowed size..bummer!
So, in order to get only the raw data, I had to turn to marshaling. It took some hair pulling but I finally managed to find a solution that both looked ok and was quite easy to understand. A lot of creds to Mike James for writing an excellent article on the topic.
For those of you who don't have strict requirements on the data size, use the BinaryFormatter. It is easy to use and does its job well. For those of you that are in a situation like mine. Take a good look at Mike James article mentioned above.
It turns out that many types can not be shared straight up between managed code and unmanaged code. To do this the types needs to be converted. This conversion process is called marshaling.
I got some real hands on experience with marshaling the last week when I worked on moving and re-writing C++ code to C#. One of the structs (that included an array of another type of struct, that in turn included another struct...) in the C++ code is written to disk with exact requirements on the number of bytes the data is allowed to occupy.
My first naive approach was to simply replicate the structs in C# and, using a BinaryFormatter and a MemoryStream, convert them to a byte array. This approach worked very well in converting back and forth between objects and binary data. Unfortunately for me the size of the byte array became about twice as large as the allowed size..bummer!
So, in order to get only the raw data, I had to turn to marshaling. It took some hair pulling but I finally managed to find a solution that both looked ok and was quite easy to understand. A lot of creds to Mike James for writing an excellent article on the topic.
For those of you who don't have strict requirements on the data size, use the BinaryFormatter. It is easy to use and does its job well. For those of you that are in a situation like mine. Take a good look at Mike James article mentioned above.
Kommentarer
Skicka en kommentar