current community

your communities

more stack exchange communities

Stack Exchange
tour help
stack overflow careers
Take the 2-minute tour ×
Stack Overflow is a question and answer site for professional and enthusiast programmers. It's 100% free, no registration required.
up vote 0 down vote favorite
1

How do you declare a struct in C# with a fixed sized array of another struct ? I need this declared so that it works. Or is my approach wrong if I want the bitmapinfo (-header) created ?

[StructLayout(LayoutKind.Sequential)]
public struct RGBQUAD
{
    public byte b;
    public byte g;
    public byte r;
    public byte reserved;
}

[StructLayout(LayoutKind.Sequential)]
public struct BITMAPINFO
{
    public BITMAPINFOHEADER bmiHeader;
    [MarshalAs(UnmanagedType.ByValArray, ArraySubType = UnmanagedType.LPStruct, SizeConst = 1)]
    public RGBQUAD[] bmiColors;
}

Edit: What I learned myself is that UnmanagedType.ByValArray and SizeConst = 1 is important here to keep the BITMAPINFO marshal size constantly at 44 bytes event if I assign 256 sized array of RGBQUAD.

Edit 2: But SizeConst mustn't be smaller than the actual array size otherwise unmanaged code can crash the app.

share|improve this question
    
You can't declare a fixed-size struct, you can only declare how a struct gets marshalled during interop. –  Peter Ritchie Feb 25 '13 at 16:38
    
Maybe this will help? pinvoke.net/default.aspx/Structures/BITMAPINFOHEADER.html –  Peter Ritchie Feb 25 '13 at 16:39

2 Answers 2

up vote 2 down vote accepted

As noted in the comments, in C# you can't create a fixed size array of structs in a struct.

But that's not what you want. In the declaration

typedef struct tagBITMAPINFO {
    BITMAPINFOHEADER bmiHeader;
    RGBQUAD          bmiColors[1];
} BITMAPINFO, *PBITMAPINFO;

bmiColors is declared as a fixed-length array but it is really a place holder for a variable length array of RGBQUAD. The actual length of the array depends (in a slightly complicated way) on the value of bmiHeader.biClrUsed.

How you handle this in C# depends entirely on what you are doing with the BITMAPINFO structure.

Update

I've just seen your other question. You're passing the BITMAPINFO to SetDIBits and your colour table always has 256 entries. So in the declaration of BITMAPINFO.bmiColors set SizeConst to 256.

share|improve this answer
    
I need a reusable solution for any palette size. So I would like to keep the size original at 1. Does this mean my declaration here is correct ? –  Bitterblue Feb 26 '13 at 7:41
    
When you marshal a structure from .Net into a native API the data is actually copied. If you set SizeConst to 1, then only one entry of your palette will be copied. At best, your image colours will be wrong. At worst, SetDIBits will crash when it reads past the end of the allocated buffer. If you set it to 256 you don't have to use all the entries. –  arx Feb 26 '13 at 14:06
    
Yes, this fixes the crash at least at CreatePalette. Can I make Marshal copy dynamically in case I don't have 256 colors but 16 or 8 ? –  Bitterblue Feb 26 '13 at 14:22
    
Structs don't have a dynamic length, so the marshaller doesn't support this as a built-in option. You can always write your own marshalling code (using the methods in the Marshal class) but doing this on each call to SetDIBits might be slower than using the built-in support. Copying 256 RGBQUADs doesn't take a long time. –  arx Feb 26 '13 at 14:33
    
But you could make it faster. Presumably the information passed in BITMAPINFO (i.e. bitmap dimensions and palette) don't change between calls. So you could marshal the structure once into unmanaged memory and use this every time you call SetDIBits. But I really wouldn't bother making any changes until you'd profiled your code and determined that the overhead of marshalling BITMAPINFO was actually significant. –  arx Feb 26 '13 at 14:35
up vote 3 down vote

Unfortunately (and IMHO somewhat surprisingly given that one of the design objectives of .net was to facilitate interop with COM) the .net runtime has no understanding of any kind of array other than standalone System.Array objects or the character arrays built into System.String. Although C# offers fixed array types, those are manipulated using pointer arithmetic in a way that the .net runtime doesn't really understand and cannot validate.

It is possible to create something that behaves like a fixed-sized array by defining a struct with elements a0, a1, a2, etc. and writing an indexed property which uses a switch statement or other such construct to access elements thereof, but such constructs are likely to perform much slower than normal arrays.

If you don't need to use safe code (and given the fact that you're interoperating with what's probably untrusted managed code, "safe" code probably isn't a consideration) I'd suggest that you define your RGBQUAD using explicit layout, to overlap an int with your other data, and then having your other structure include a fixed int[whateverSize]; to hold the RGB data. Its indexed accessor should be able to convert int to RGBQUAD rather efficiently by just reading/writing its "int" member.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.