File written by old C code, moved to C#

Question

I'm working on converting an old C program (currently run on UNIX) into our C# system.

The program builds some data into several structs and then writes a file using a series of fwrites like this:

fwrite ( &loc,  sizeof ( struct loc_items ),  1, hdata.fp );

With loc being the data, struct loc_items being the struct it's a type of.

My question is if I will be able to match this file output with C#? Assuming I can match the endian format and sizes of floats, ints, etc.

Since it's writing the entire struct in the fwrite, I don't know the order that it's writing the different variables of the struct. Also is there any issue with padding around the struct the fwrite might put?

Any advice?

Answer 1

Make a quick unit test that generates a file using the ported C# version and then does a compare of the contents from a file generated by the system. Then you will at least have a test that discovers discrepancies automatically.

Answer 2

You have a variety of options.

• Use BinaryWriter and write each field of the struct individually.
• Use marshalling to get an array of bytes for the struct and write that to your stream.
• Use a library like Miguel de Icaza's struct library.

Personally, I dislike the use of marshalling in this context. I would simply add some "Open/Save", "Read/Write", "Hydrate/Dehydrate", "Serialize/Deserialize" methods to your struct or class.

class POCO {
    int aaa;
    double bbb;

    public void Read(BinaryReader r) {
        aaa = r.ReadInt32();
        bbb = r.ReadDouble();
    }
    public void Write(BinaryWriter w) {
        w.Write(aaa);
        w.Write(bbb);
    }
}

This way, you're in full control of what gets read and written.