Take the 2-minute tour ×
Programmers Stack Exchange is a question and answer site for professional programmers interested in conceptual questions about software development. It's 100% free, no registration required.

Python's lack of pointers makes data-structure construction intuitively more challenging, and has so much built in functionality that, as an amateur, I can't see when, why, or how you would create something beyond what's already there.

When have you built your own, and why did you do it? E.g. special types of trees, etc.

share|improve this question
4  
You create a custom data structure when you need it and Python doesn't provide it out of the box. en.wikipedia.org/wiki/List_of_data_structures –  Robert Harvey Dec 11 '12 at 19:39
9  
"Python's lack of pointers makes data-structure construction intuitively more challenging" Uhm, no it doesn't. It's roughly the same as in C# or Java. –  user16764 Dec 11 '12 at 19:47
    
The whole field of object-orientation is about creating custom data structures - in most object oriented languages they are called "classes". So you actually ask "shall anyone create a new class, there are enough classes already there". Hope this makes the absurdity of your question clear. –  Doc Brown Dec 11 '12 at 19:51
    
@DocBrown: Probably pedantic but: the whole field of programming is all about creating custom data structures (with algorithms to operate on them)... OOP just commonly uses objects as the primary mechanism to do that. I think that knowing this increases the clarity you mention. –  Steve Evers Dec 11 '12 at 20:31
    
It sounds like your hang-up is about pointing to things. Most languages without explicit pointers still support references (assigning any object = to another that isn't a value type). Generally you just create a containing data, and a reference to another node, or list of nodes, or two nodes, or however your structure is going to be represented. Or write it in C and link to it. –  KChaloux Dec 14 '12 at 15:34
add comment

2 Answers 2

up vote 2 down vote accepted

Well, that kind of depends on what you are willing to call a "data structure." According to Wikipdia, a data structure is simply a "particular way of storing and organizing data in a computer so that it can be used efficiently." So, therefore, classes would be a form of data structure, and classes are very widely used in Python. But, for the sake of this answer, I will assume you are more interested in what you might learn about in a data structures and algorithms class (e.g. trees, linked lists, queues, hashes, etc...).

Now, because of Python's pseudo-code-like syntax it can be a very useful language for implementing data structures in. If for no other purpose than just to aid in understanding these basic concepts. For example, when I first learned about linked list I decided to implement them in Python:

class node:
    def __init__(self, data = None, is_header = False):
        self.data = data
        self.next = None
    def __str__(self):
        return "|{0}| --> {1}".format(str(self.data), str(self.next))


class linked_list:
    def __init__(self):
        self.header = node("!Header!", True)

    def add_node(self, add_me):
        current_node = self.header
        while 1:
            if current_node.next == None:
            current_node.next = add_me
            break
            else:
            current_node = current_node.next    

    def print(self):
        print (str(self.header))

Now, this isn't a perfect example, nor is it even a totally proper implementation of a linked list, but it goes to illustrate something:

Python's simple syntax can be helpful in understanding data structures

Another example would be a priority queue that I built back in the day:

class priority_queue:
    def __init__(self):
        self.element = []
        self.priority = []

    def enqueue_with_priority(self, me, priority):
        where = 0
        for i in range(len(self.element)):
            if not priority < self.priority[i]:
            where = i
            break

        self.element.insert(where, me)
        self.priority.insert(where, priority)

    def dequeue_highest(self):
        return self.element.pop(0)

    def print(self):
        for i in self.element:
            print i

Again, not a perfect implementation but it illustrates another benefit of coding a data structure in Python:

Python is useful in prototyping data structures to optimize them for lower-level programming languages

Looking back on this code I see flaws in my implementation. The Python code, however, tends to be short and sweet. So if I wanted to implement a data structures in a lower-level language (such as a c-style language) I could first generate a quick Python prototype and then optimize later.

Finally, I think Python can help in the development of data structures, because:

In Python, development is quick, mistakes are allowed and you can try things out.

Imagine you are building a hash-table-like data structure, in a strongly-typed, compiled language you would usually try things out in an IDE, then have to compile and run it. In Python, you can just pull up IDLE, iPython or the Python interpreter and just try things out! No need to recompile for each little change to the hash function you want to try -- just plug it into the interpreter.

So, in conclusion, I guess what I'm saying is that I agree with you: there's not a lot of practicality in building your own data structures (since most anything you may want has already been implemented and optimized). However, for me, there is a lot of educational benefit (because of Python's ease of syntax), as well as a lot of creative and developmental freedom (due to Python's low-constraint, EAFP design).

It is important to note that although python (through its wide-reaching library) provides many standard data structures, a "data structure" (by definition) can be almost anything. So, in Python as well as any other language we may use to solve non-trivial problems, we are going to need to define new data structures. Therefore, it is quite arguable that serious Python developers create custom data structures just as much as serious developers in other languages do.

share|improve this answer
    
Are you actually quoting someone or is this just for presentation? –  phant0m Dec 11 '12 at 20:14
    
@phant0m, just for presentation, should I change it? –  mjgpy3 Dec 11 '12 at 20:15
3  
It's a little confusing, yes. Also, there are plenty of custom data structures you'll need from time to time, a data structure is just structured data; the classic structures are indeed already covered with great implementations, but don't think you ever are done building more. –  Martijn Pieters Dec 11 '12 at 20:20
    
@MartijnPieters, thanks for the insight. –  mjgpy3 Dec 11 '12 at 20:32
add comment

To make data structures in Python you'd use a class. References in Python (and other reference based languages) act like pointers and so there is no restriction to being able to implement a data structure.

Python does have many data structures builtin. But then so do various other languages. See the STL for C++ or Java's standard library. In both C++ and Java you also rarely need to implement your own data structure because most of the basic structures already exist.

But sometimes, there is a data structure you need that Python doesn't have builtin. For example, python doesn't have any builtin tree-based structures, or linked lists, etc. You don't need these all that often, but when you need them, you need them.

share|improve this answer
2  
A more common (due to its generality) data structure not in the Python standard library is a graph. This is partially because (as python-ideas recently discussed) there are so many radically different uses of them that a single implementation can't possibly cover most use cases. I regularly create graphs, though they're rarely explicitly treated as such. –  delnan Dec 11 '12 at 20:16
    
@delnan, that's true. We create graphs all the time even without thinking about it. –  Winston Ewert Dec 12 '12 at 15:56
add comment

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.