Does a StringBuilder initialized with a string contain exactly (only) enough space for that string?

Question

I'm wondering if this code ...

StringBuilder sb = new StringBuilder("Please read the following messages.");

... initializes sb with a buffer exactly as large as the string passed to the constructor. On the one hand, this would seem the most logical thing. On the other hand, it seems to kind of defeat the purpose of the StringBuilder class for one of its most common uses, which is to provide mutability to make repeated appends more efficient. (The very first call to Append, if the answer to my question is "yes", would require sb to resize itself.)

Then again, I suppose one could view this as analogous to the constructor for List<T> that takes an IEnumerable<T> as a parameter. Maybe the assumption in this case is that you're not planning on appending a lot, but rather on manipulating what's already there.

The only real research I've done on this was to check the MSDN documentation on StringBuilder, which didn't provide an answer (it says the constructor initializes the instance "using the specified string," but doesn't indicate how the string is used).

---

EDIT: So it's "implementation-specific"... does this not seem weird to anyone else? I mean, the purpose of the StringBuilder class is to offer an alternative to performing a lot of operations on a string, creating a ton of immutable string instances along the way; therefore, it's for efficiency. I feel like the behavior of this constructor ought the be specified, so that the developer can make an informed decision how to use it regardless of platform.

I mean, it is implemented by Microsoft a certain way; they could easily have put that in the documentation (forcing other implementations to follow suit). Just a personal source of puzzlement...

Answer 1

It's an implementation detail that you shouldn't need to worry about. However, using .NET reflector, and looking in the (string,int32,int32,int32) overload of the constructor (which the other constructors call), we can see that it picks a capacity that is a multiple of 16 (next largest over the requested size)

Edit

Actually, it's 16 x 2^n, with the value of "n" selected to be the next largest size

Answer 2

Check the StringBuilder's Capacity member.

From MSDN:

The StringBuilder dynamically allocates more space when required and increases Capacity accordingly. For performance reasons, a StringBuilder might allocate more memory than needed. The amount of memory allocated is implementation-specific.

Answer 3

The constructor you linked to is probably chained to the StringBuilder(String, Int32, Int32, Int32):

public StringBuilder(
  string value,
  int startIndex,
  int length,
  int capacity
)

So, for the string, it would probably pass through: string, 0, string.Length, string.Length. Or something similar that makes sense in the StringBuilder context.

Answer 4

The constructor that eventually gets called is:

// "Please read the following messages.".Length = 35
public StringBuilder(string value, int startIndex, int length, int capacity)
public StringBuilder("Please read the following messages.", 0, 
        "Please read the following messages.".Length, 16)

(This is nothing that the other answers don't provide, and is just from reflector)
If the capacity is less than the length of the string, which it is in this case:

while (capacity < length)
{
    capacity *= 2;
    if (capacity < 0)
    {
        capacity = length;
        break;
    }
}

In Mono, the StringBuilder(string val) constructor allocates the capacity to int.MaxValue until an append occurs.

The real answer lies in the method that ends up being called internally in the CLR, where length is the capacity:

[MethodImpl(MethodImplOptions.InternalCall)]
private static extern string FastAllocateString(int length);

I can't find the source for this in the SSCLI however the Mono version (\mono\metadata\object.c) does it like this:

mono_string_new_size (MonoDomain *domain, gint32 len)
{
    MonoString *s;
    MonoVTable *vtable;
    size_t size = (sizeof (MonoString) + ((len + 1) * 2));

...
}

Which is the size in bytes of a MonoString object, plus the length times 2.