checksum theory? - Dev Shed

Dev Shed Forums Sponsor:

November 26th, 2003, 10:19 PM

theRaginCajun

Contributing User

Join Date: Feb 2003

Location: USA

Posts: 144

Time spent in forums: < 1 sec
Reputation Power: 9

checksum theory?

Hi everyone,

I was trying to put some error checking in this file serialization I was doing. I was wondering about how checksum works - is it basically like you add up the values of all the variables you're serializing then store that in one variable and also write that to the same file - then when you read it back in, if all the read in varaibles' values don't equal the checksum value from the last save, then you know the file is corrupt / messed up?

Code:

WriteFile()
{
    int         x = 5;
    int         y = 10;
    CString z = "hello";
    
    int nCheckSum = (x + y + ASCII value of 'hello'?);

    WriteTheFile();
}

ReadFile()
{
    int         x, y, z, nCheckSum;
    CString z;

    ar >> x;
    ar >> y;
    ar >> z;
    ar >> nCheckSum;

    if (x + y + ASCII value of z? != nCheckSum)
        return false;
}

Thanks!

November 26th, 2003, 11:14 PM

peenie

Google Relay Server

Dev Shed Intermediate (1500 - 1999 posts)

Join Date: Oct 2003

Location: Oh christ I don't even know any more.

Posts: 1,810

Folding Points: 9953 Folding Title: Novice Folder

Time spent in forums: 2 Weeks 1 Day 19 h 12 m 24 sec
Reputation Power: 436

Yeah that's about it. There's lots of different checksum algorithms. CRC16 and CRC32 are fairly common: http://www.vbaccelerator.com/home/V...C32/article.asp .

If simplicity is all you care about, you could just add up the ASCII values of all the characters. This is generally not very reliable though, because swapping two characters will yield the same checksum, and so will say, changing a 'b' to a 'c' and a 'x' to a 'w' in the same file. The MD5 hashing algorithm is very reliable, and you can find implementations of it all over the place; but it's probably a bit of overkill.

That's the general idea though. Compute checksum, store somewhere. Then, when you read the data back in, compute checksum, compare it with stored checksum. Checksums are useful for detecting data corruption because the chance of both the data being corrupted AND the checksum being accidently corrupted to the "correct" value are very slim indeed.

November 27th, 2003, 02:36 AM