I'm currently writing a program that (as of right now) utilizes about five linked lists, and there's about 50 functions that modify or otherwise use those linked lists. I am somewhat torn on how exactly the head node should be implemented.
Implementation #1 Using a dummy head node
If I create a dummy head node, the headNode never has to be deleted, even if there is no data for that linked list to represent. Below are two different possible linked lists showing this idea.
Format: (object)--name_of_pointer_to_next_object--> (next object)
(headNode)--headNode.pNext--> (Nothing/NULL)
(headNode)--headNode.pNext--> (dataNode1)--dataNode1.pNext--> (dataNode2)--dataNode2.pNext--> (Nothing/NULL)
The benefit of this implementation is that, for functions that work with the linked lists, they don't have to have special code for the case where the head node is Nothing or (in c++), NULL. Below is an example of appending a node onto the end of the linked list using this implementation.
Public Function AppendNode(theLinkedList as NODE_, theNewNode as NODE_)
dim lastNode as NODE_
Set lastNode = theLinkedList
Do While Not lastNode.pNext Is Nothing
Set lastNode = lastNode.pNext
Loop
Set lastNode.pNext = theNewNode
End Function
Implementation #2 headNode can be deleted
If I allow the headNode to be deleted, this poses problems and increases the volume of code that I have to write. Below is the same set of data, except this time the head is a legitimate node that contains data.
Format: (object)--name_of_pointer_to_next_object--> (next object)
(Nothing/NULL)
(dataNode1)--dataNode1.pNext--> (dataNode2)--dataNode1.pNext--> (Nothing/NULL)
And here is the same function, except that this time it has to beware of the possibility that the head node is Nothing/NULL.
Public Function AppendNode(theLinkedList as NODE_, theNewNode as NODE_)
dim lastNode as NODE_
If theLinkedList Is Nothing Then
Set theLinkedList = theNewNode
Else
Set lastNode = theLinkedList
Do While Not lastNode.pNext Is Nothing
Set lastNode = lastNode.pNext
Loop
Set lastNode.pNext = theNewNode
End If
End Function
Obviously, I am leaning towards Implementation #1. It requires at least 4 less lines of code each time I want to use the linked lists (considering the fact that I will do this hundreds of times, you can multiply that 4 lines by 300 or so; e.g. it will save me from writing 1,200 lines of code), AND perhaps most importantly, it will cut down on the amount of state my program has. All of the possible states that my program will be in will only require me to look for pNext Is Nothing, and reducing the amount of state is greatly appreciated at this point because this program is going to be a monster that will end up being around 20k lines of rather complicated code that has to deal with a lot of state.
Am I wrong in thinking that Implementation #1 is the best way to do this? I can't come up with a single reason why Implementation #2 would be superior.
It is very common (though not required) to have a special, reserved node at the start and end of a linked list. It is typically called a sentinel node.
As an aside, I suggest taking the "about 50 functions that modify or otherwise use those linked lists" and moving them to a Class that encapsulates them.
You should not have code that looks like this:
Dim theLinkedList as NODE_
Set theLinkedList = New NODE_
'bunch of work with the list
'bunch of work with the list
'bunch of work with the list
Dim theNewNode as NODE_
Set theNewNode = New NODE_
Set theNewNode.Next = someDataObject
Call AppendNode(theLinkedList, theNewNode)
Instead the code should just look like this:
Dim theList As LinkedList
Set theList = New LinkedList
'etc
Call theList.Append(someDataObject)
See the difference? The LinkedList calss wraps and hides the details. It can handle things like whether there is a sentinel node at the head or tail as well so that all the rest of your project's code is blissfully unaware.