storage space of strings

vik · #1 (**permalink**) July 2nd, 2009, 03:32 PM

Greetings to everybody!

As far as I know a string needs as much memory as many character it contains.
Opposing this theory I have two - theoretically equivalent - ways of building a string, but the storage space seems to be very different. Here is the code:

Code:

    class Program
    {
        static void Main(string[] args)
        {
            A a = new A();

            BinaryFormatter bf = new BinaryFormatter();
            FileStream fs = new FileStream(path, FileMode.Create, FileAccess.ReadWrite);
            bf.Serialize(fs, a);
            fs.Close();
        }
    }

    [Serializable]
    class A
    {
        List<string> aha;
        public List<string> Aha
        {
            get { return aha; }
        }
        public A()
        {
            aha = new List<string>();
            for (int i = 0; i < 10000; i++)
            {
                aha.Add("a9999b");
            }
        }
    }

and the other version:

Code:

    class Program
    {
        static void Main(string[] args)
        {
            A a = new A();

            BinaryFormatter bf = new BinaryFormatter();
            FileStream fs = new FileStream(path, FileMode.Create, FileAccess.ReadWrite);
            bf.Serialize(fs, a);
            fs.Close();
        }
    }

    [Serializable]
    class A
    {
        List<string> aha;
        public List<string> Aha
        {
            get { return aha; }
        }
        public A()
        {
            aha = new List<string>();
            for (int i = 0; i < 10000; i++)
            {
                aha.Add("a" + i.ToString() + "b");
            }
        }
    }

The only difference is the way the string is built, and all the strings in the first version are non-longer than the ones in the second, so I thought that file will occupy less bytes. But surprisingly the first file's size is 50kB and the second's is 117kB.
Could You tell me what's wrong? Are my starting hypothesys' wrong or is this normal, and if not how should the second solution occupy the same amount of memory (or less, as it's strings are shorter?)?
Thank You in advance!
Vik

Lee Dumond · #2 (**permalink**) July 2nd, 2009, 07:16 PM

This is because of the way the BinaryFormatter's Serialize method works.

When a series of identical bytes is serialized using the BinaryFormatter, it doesn't just repeat the whole sequence over and over. It optimizes by using the format to indicate that the same sequence of bytes is repeated.

Since the first stream consists of "a9999b" over and over, a lot of space is saved there.

In the second stream, each sequence of bytes is unique; therefore, the formatter cannot optimize in the same fashion, and so the stream ends up being much larger.

Make sense?

vik · #3 (**permalink**) July 4th, 2009, 11:59 AM

Thank You for Your reply, this way it's clear now! Thank You!

Need to download code?

Navigation