The Big Programming Thread - Page 631

Prev 1 629 630 631 632 633 1031 Next

Thread Rules
1. This is not a "do my homework for me" thread. If you have specific questions, ask, but don't post an assignment or homework problem and expect an exact solution.
2. No recruiting for your cockamamie projects (you won't replace facebook with 3 dudes you found on the internet and $20)
3. If you can't articulate why a language is bad, don't start slinging shit about it. Just remember that nothing is worse than making CSS IE6 compatible.
4. Use [code] tags to format code blocks.

solidbebe

Netherlands4921 Posts

May 13 2015 21:05 GMT

#12601

double post

Ben...

Canada3485 Posts

May 13 2015 21:12 GMT

#12602

On May 14 2015 05:49 killa_robot wrote:
You don't go to school to learn how to work though, you go to school to learn how to learn. One of the greatest disconnects we have right now is between schools and employers.

Yup. I've seen this topic discussed to death on a different forum and this is what it always comes down to. Is university supposed to be a place of learning or a place to prepare you for working? The business types seem to think it should prepare you for working in the field while the academic types feel it should teach you the underlying concepts and more of the "why's" than "hows" and that in knowing these concepts, you can be a more adaptable programmer since you understand things in general and can apply them well and learn quickly.

iaretehnoob

Sweden741 Posts

May 13 2015 22:37 GMT

#12603

On May 14 2015 06:04 solidbebe wrote:
So Im very new to C++ and don't know anything about the language's intricacies yet. I've implemented this algorithm which computes the sum of values on the even levels of a tree. Given that I have to use lists (and not vectors or what have it), is this a good implementation?

+ Show Spoiler +


	void visitTreeEvenLevels(std::list<Tree> treelist, int i, float& total){
		if (treelist.size() == 0){
			return;
		}
		std::list<Tree>::const_iterator ptr = treelist.begin();
		std::list<Tree> cont;
		for (; ptr != treelist.end(); ptr++){
			std::list<Tree> temp = ptr->children;
			std::list<Tree>::const_iterator ptr2 = temp.begin();
			for (; ptr2 != temp.end(); ptr2++){
				cont.push_front(*ptr2);
				if (i % 2 == 0){
					total += ptr2->value;
				}
			}
		}
		++i;
		visitTreeEvenLevels(cont, i, total);
	}

Just looking for criticism from experienced C/C++ people

Try changing your function signature to something like

void visitTreeEvenLevels(std::list<Tree>::const_iterator treelist, int i, float& total)

and doing it without creating temporary lists. Other than that it seems fine, but maybe clean up the naming a bit, stuff like ptr and ptr2 or using i for anything but a simple loop counter.

Shield

Bulgaria4824 Posts

May 14 2015 00:09 GMT

#12604

On May 14 2015 07:37 iaretehnoob wrote:

Show nested quote +

Try changing your function signature to something like

void visitTreeEvenLevels(std::list<Tree>::const_iterator treelist, int i, float& total)

and doing it without creating temporary lists. Other than that it seems fine, but maybe clean up the naming a bit, stuff like ptr and ptr2 or using i for anything but a simple loop counter.

I hope it works as I can't test it but here is some slight improvement:


void visitTreeEvenLevels(std::list<Tree> treelist, int& i, float& total)
{
    if (treelist.empty()) {
        return;
    }

    std::list<Tree> cont;

    for (const Tree& parent : treelist)
    {
        for (const Tree& child : entry.children)
	{
	    cont.push_front(child);
	    if (i % 2 == 0){
	        total += child.value;
	    }
	}
    }
    ++i;
    visitTreeEvenLevels(cont, i, total);
}

1. empty() is better than size() == 0
2. for-each loop seems nicer if your compiler supports C++11
3. Edit: I've changed method's signature to be int& instead of int. You can either ++i directly or ++i before the recursive method, then pass i. Both should work. This should avoid unnecessary copies of 'i'. I guess the latter is more readable and doesn't rely on the reader to know that pre-increment returns a reference.

Edit: You can also avoid copies as suggested by iaretehnoob.

Cyx.

Canada806 Posts

May 14 2015 00:36 GMT

#12605


	void visitTreeEvenLevels(std::list<Tree> treelist, int i, float& total){
		if (treelist.size() == 0){
			return;
		}
		std::list<Tree>::const_iterator ptr = treelist.begin();
		std::list<Tree> cont;
		for (; ptr != treelist.end(); ptr++){
			std::list<Tree> temp = ptr->children;
			std::list<Tree>::const_iterator ptr2 = temp.begin();
			for (; ptr2 != temp.end(); ptr2++){
				cont.push_front(*ptr2);
				if (i % 2 == 0){
					total += ptr2->value;
				}
			}
		}
		++i;
		visitTreeEvenLevels(cont, i, total);
	}

Just looking for criticism from experienced C/C++ people

Looks okay, I would really prefer range-based for loops if you can use them:

for (auto ptr : treelist)

And

treelist.size() == 0

is called

treelist.empty()

I'm not sure about that 'i' you're using to track your depth, either - would it be maybe clearer/easier to just skip a level instead of checking which level you're on every time? Something along the lines of

//visitTreeEvenLevels(cont, i, total)
for (auto it : cont) {
    visitTreeEvenLevels(it->children, total);
}

Dunno though.

Shield

Bulgaria4824 Posts

May 14 2015 00:40 GMT

#12606

On May 14 2015 09:36 Cyx. wrote:

Show nested quote +

Looks okay, I would really prefer range-based for loops if you can use them:

for (auto ptr : treelist)

And

treelist.size() == 0

is called

treelist.empty()

//visitTreeEvenLevels(cont, i, total)
for (auto it : cont) {
    visitTreeEvenLevels(it->children, total);
}

Dunno though.

auto is kind of bad because it's not explicit, but you may also call the copy constructor. Try auto& instead of auto.

Khalum

Austria831 Posts

May 14 2015 00:56 GMT

#12607

On May 14 2015 07:37 iaretehnoob wrote:

Show nested quote +

Try changing your function signature to something like

void visitTreeEvenLevels(std::list<Tree>::const_iterator treelist, int i, float& total)

and doing it without creating temporary lists. Other than that it seems fine, but maybe clean up the naming a bit, stuff like ptr and ptr2 or using i for anything but a simple loop counter.

You can go one step further and pass the iterator as a reference:

void visitTreeEvenLevels(std::list<Tree>::const_iterator& treelist, int i, float& total)

RoyGBiv_13

United States1275 Posts

May 14 2015 05:00 GMT

#12608

On May 14 2015 09:40 darkness wrote:

Show nested quote +

On May 14 2015 09:36 Cyx. wrote:


	void visitTreeEvenLevels(std::list<Tree> treelist, int i, float& total){
		if (treelist.size() == 0){
			return;
		}
		std::list<Tree>::const_iterator ptr = treelist.begin();
		std::list<Tree> cont;
		for (; ptr != treelist.end(); ptr++){
			std::list<Tree> temp = ptr->children;
			std::list<Tree>::const_iterator ptr2 = temp.begin();
			for (; ptr2 != temp.end(); ptr2++){
				cont.push_front(*ptr2);
				if (i % 2 == 0){
					total += ptr2->value;
				}
			}
		}
		++i;
		visitTreeEvenLevels(cont, i, total);
	}

Just looking for criticism from experienced C/C++ people

Looks okay, I would really prefer range-based for loops if you can use them:

for (auto ptr : treelist)

And

treelist.size() == 0

is called

treelist.empty()

//visitTreeEvenLevels(cont, i, total)
for (auto it : cont) {
    visitTreeEvenLevels(it->children, total);
}

Dunno though.

auto is kind of bad because it's not explicit, but you may also call the copy constructor. Try auto& instead of auto.

auto isn't bad, though it is not explicit, which can be bad.

For the pros, It allows the compiler to take a lot of leniency on the data type when trying to optimize, it reduces work in refactoring types through multiple abstraction layers, and it still provides static type checking.

For the cons, it's not explicit and it even though you probably know the type at the time of you originally write the declaration of the variable, you may forget what it should be and misuse the variable later.

For all the benefits without the drawbacks, if you have a long function or a variable with complicated context, on the same line as the declaration or in the function comment block, mention the expected type and usage of those variables, even if you end up being wrong about the type the compiler uses. Later in the function, use assertions to force the type the what you expect, i.e, if "auto x" is an unsigned integer under MAX_X, Assert(x>0 && x < MAX_X);
+ Show Spoiler +

you ARE using assertions, right?

I personally disapprove of ever using auto for floats or doubles. If you're using a float or a double, you should be extra careful with that type.

netherh

United Kingdom333 Posts

May 14 2015 06:09 GMT

#12609

On May 14 2015 14:00 RoyGBiv_13 wrote:
I personally disapprove of ever using auto for floats or doubles. If you're using a float or a double, you should be extra careful with that type.

Why?

I've been following the "almost always auto" advice for a while now, and the only time I've had any trouble is with a library that uses expression templates.

Mstring

Australia510 Posts

May 14 2015 07:44 GMT

#12610


	void visitTreeEvenLevels(std::list<Tree> treelist, int i, float& total){
		if (treelist.size() == 0){
			return;
		}
		std::list<Tree>::const_iterator ptr = treelist.begin();
		std::list<Tree> cont;
		for (; ptr != treelist.end(); ptr++){
			std::list<Tree> temp = ptr->children;
			std::list<Tree>::const_iterator ptr2 = temp.begin();
			for (; ptr2 != temp.end(); ptr2++){
				cont.push_front(*ptr2);
				if (i % 2 == 0){
					total += ptr2->value;
				}
			}
		}
		++i;
		visitTreeEvenLevels(cont, i, total);
	}

Just looking for criticism from experienced C/C++ people

Keep in mind exactly what happens when you say:


std::list<Tree> temp = ptr.children;

This will duplicate the entire children list, including all the children's children, and their children, etc.. Since you simply want to iterate over the whole thing and sum some values, what reason do you have for wanting to make a copy of anything? Your 'cont' list behaves similarly. Finally you're passing the list by value into the function, resulting in even more copies every time your function is called.

Passing references around makes avoiding this kind of thing easy:


void SumTreeEvenLevels(std::list<tree_node> &TreeList, int Level, float &Total)
{
  for(std::list<tree_node>::iterator Node = TreeList.begin(); Node != TreeList.end(); Node++)
  {
    if(Level % 2 == 0)
      Total += Node->Value;

    SumTreeEventLevels(Node->Children, Level + 1, Total);
  }
}

Blisse

Canada3710 Posts

May 14 2015 08:12 GMT

#12611

void SumTreeEvenLevels(
    std::list<tree_node*>::const_iterator& current, 
    std::list<tree_node*>::const_iterator& end, 
    int level, 
    float& total
)
{
    while (current != end)
    {
        if (level % 2)
        {
            total += (*current)->value;
        }

        SumTreeEvenLevels(
            (*current)->children.begin(), 
            (*current)->children.end(), 
            level + 1, 
            total
        );
    
        ++current;
    }
}

float SumTreeEvenLevels(
    tree_node& root
)
{
    float value = root.value;
    int level = 1;

    SumTreeEvenLevels(
        root.children.begin(), 
        root.children.end(), 
        level, 
        value
    );

    return value;
}

lalalala probably this would be satisfactory

someone should summon nunez

Khalum

Austria831 Posts

May 14 2015 09:28 GMT

#12612

On May 14 2015 17:12 Blisse wrote:

[..]
if (level % 2)
[..]

Nope, that's true for all odd levels.

[edit] But you could start counting with 0...

netherh

United Kingdom333 Posts

May 14 2015 10:12 GMT

#12613

Old versions

+ Show Spoiler +


#include <vector>
#include <list>
#include <functional>

struct Tree
{
	Tree(float value): Value(value) { }

	float Value;

	std::vector<Tree> Children;
};

float AccumulateEven(std::list<Tree> const& trees)
{
	typedef std::function<void(Tree const& tree, float& result)> FunctionT;

	auto even = FunctionT();
	auto odd = FunctionT();

	even = [&odd] (Tree const& tree, float& result)
	{
		result += tree.Value;

		for (auto const& child : tree.Children)
			odd(child, result);
	};

	odd = [&even] (Tree const& tree, float& result)
	{
		for (auto const& child : tree.Children)
			even(child, result);
	};

	auto result = 0.f;

	for (auto const& tree : trees)
		even(tree, result);

	return result;
}

Yay, lambdas!

EDIT: Actually, something like the following works quite nicely:


#include <vector>
#include <list>

struct Tree
{
	Tree(float value): Value(value) { }

	float Value;

	std::vector<Tree> Children;
};

template<bool IsEven>
void AccumulateEven(Tree const& tree, float& result)
{
	if (IsEven)
		result += tree.Value;

	for (auto const& child : tree.Children)
		AccumulateEven<!IsEven>(child, result);
};

float AccumulateEven(std::list<Tree> const& trees)
{
	auto result = 0.f;

	for (auto const& tree : trees)
		AccumulateEven<true>(tree, result);

	return result;
}

I expect the if (IsEven) branch will get optimized out entirely for the AccumulateEven<false> instantiation.

EDIT: Wait... aren't we just doing this? (replace vector w/ list as appropriate):


#include <vector>

struct Node
{
	Node(float value): Value(value) { }

	float Value;

	std::vector<Node> Children;
};

template<bool IsEven>
float AccumulateEven(Node const& node)
{
	auto result = 0.f;

	for (auto const& child : node.Children)
		result += AccumulateEven<!IsEven>(child);

	return IsEven ? result + node.Value : result;
};

Manit0u

Poland17196 Posts

May 14 2015 11:14 GMT

#12614

On May 14 2015 05:49 killa_robot wrote:

Show nested quote +

Nope. Used Hadoop, but apart from that didn't use any frameworks.

You don't go to school to learn how to work though, you go to school to learn how to learn. One of the greatest disconnects we have right now is between schools and employers.

Well, if you want to be an academic then I believe studying maths or philosophy would be more beneficial. CS should be pretty much purely technical (technical in learn how it's done in the real world sense) studies that prepare you for work in the industry.

Also, quite a large number of developers aren't even schooled in it. Here are some interesting resources:

http://stackoverflow.com/research/developer-survey-2015#profile-education
http://blog.codinghorror.com/why-cant-programmers-program/

There was also a great paper showing that people without CS education tend to be better programmers most of the time.

spinesheath

Germany8679 Posts

May 14 2015 11:40 GMT

#12615

On May 14 2015 19:12 netherh wrote:
EDIT: Wait... aren't we just doing this? (replace vector w/ list as appropriate):


#include <vector>

struct Node
{
	Node(float value): Value(value) { }

	float Value;

	std::vector<Node> Children;
};

template<bool IsEven>
float AccumulateEven(Node const& node)
{
	auto result = 0.f;

	for (auto const& child : node.Children)
		result += AccumulateEven<!IsEven>(child);

	return IsEven ? result + node.Value : result;
};

Templates, why are you so awesome?

On May 14 2015 20:14 Manit0u wrote:
There was also a great paper showing that people without CS education tend to be better programmers most of the time.

That sounds a lot like a newspaper headline that takes the result of the paper out of context.

Most likely the actual result is that self taught programmers are better than people who just did the mandatory stuff for their CS education. CS education doesn't make you a good programmer (at all), but if you are passionate about programming, you will be good regardless of having a CS education or not.

iaretehnoob

Sweden741 Posts

May 14 2015 12:49 GMT

#12616

On May 14 2015 09:56 Khalum wrote:

Show nested quote +

On May 14 2015 07:37 iaretehnoob wrote:


	void visitTreeEvenLevels(std::list<Tree> treelist, int i, float& total){
		if (treelist.size() == 0){
			return;
		}
		std::list<Tree>::const_iterator ptr = treelist.begin();
		std::list<Tree> cont;
		for (; ptr != treelist.end(); ptr++){
			std::list<Tree> temp = ptr->children;
			std::list<Tree>::const_iterator ptr2 = temp.begin();
			for (; ptr2 != temp.end(); ptr2++){
				cont.push_front(*ptr2);
				if (i % 2 == 0){
					total += ptr2->value;
				}
			}
		}
		++i;
		visitTreeEvenLevels(cont, i, total);
	}

Just looking for criticism from experienced C/C++ people

Try changing your function signature to something like

void visitTreeEvenLevels(std::list<Tree>::const_iterator treelist, int i, float& total)

and doing it without creating temporary lists. Other than that it seems fine, but maybe clean up the naming a bit, stuff like ptr and ptr2 or using i for anything but a simple loop counter.

You can go one step further and pass the iterator as a reference:

void visitTreeEvenLevels(std::list<Tree>::const_iterator& treelist, int i, float& total)

I think iterators are usually passed by value, which makes a lot of sense when you think of them as "fancy pointers".

On May 14 2015 19:12 netherh wrote:
EDIT: Wait... aren't we just doing this? (replace vector w/ list as appropriate):


#include <vector>

struct Node
{
	Node(float value): Value(value) { }

	float Value;

	std::vector<Node> Children;
};

template<bool IsEven>
float AccumulateEven(Node const& node)
{
	auto result = 0.f;

	for (auto const& child : node.Children)
		result += AccumulateEven<!IsEven>(child);

	return IsEven ? result + node.Value : result;
};

Nice.

Prillan

Sweden350 Posts

May 14 2015 15:17 GMT

#12617

On May 13 2015 22:30 Ropid wrote:
There's also the complete opposite, you know. I actually regret that the academic side of "computer science" wasn't there enough.

I've looked at this here today: http://ncatlab.org/nlab/show/polynomial functor

It is related to object-oriented programming. I don't understand.

You need to give room for something like this as well, or you won't find people that want to work on it in depth. Also, if people can't even vaguely get what the stuff is about, they won't really know if this is actually an interesting topic for them or not. You need to show the basics that would enable people to start approaching those kinds of topics. I felt that was missing.

Don't expect to understand anything written on ncatlab. I've studied mathematics for 4 years, with interest in algebra and category theory, and I still can't make sense of most of the stuff posted there.

Manit0u

Poland17196 Posts

May 14 2015 16:34 GMT

#12618

So, I've decided to learn me some Java. Decided to start out with something simple, like the FizzBuzz test that some people use during job interviews.


class FizzBuzz {
    public static void main(String[] args) {
        int i = 0;
        boolean isDivisibleByThree = false;
        boolean isDivisibleByFive = false;
        String msg = "";

        while (i++ < 100) {
            isDivisibleByThree = ((i % 3) == 0);
            isDivisibleByFive = ((i % 5) == 0);

            if (isDivisibleByThree && isDivisibleByFive) {
                msg = "FizzBuzz";
            } else if (isDivisibleByThree) {
                msg = "Fizz";
            } else if (isDivisibleByFive) {
                msg = "Buzz";
            } else {
                msg = Integer.toString(i);
            }

            System.out.println(msg);
        }
    }
}

Could someone point me into right direction so that I can move away from the imperative thinking and towards full OOP and the "Java-way" (java specific of course, I know a fair bit about OOP but very little about Java)? How would you write this simple app in the way that Java should be written?

spinesheath

Germany8679 Posts

May 14 2015 17:01 GMT

#12619

If you're asking how to over-engineer this program, there is Enterprise FizzBuzz. There is no point in adding classes for a program as simple as this one.

Personally, I'd refactor your code to something like this, but that's hardly Java specific:


class FizzBuzz {
    public static void main(String[] args) {
        for (int = 0; i < 100; ++i) {
            String msg = GetMessage(i);
            System.out.println(msg);
        }
    }

    private static String GetMessage(int i)
    {
        boolean isDivisibleByThree = ((i % 3) == 0);
        boolean isDivisibleByFive = ((i % 5) == 0);

        if (isDivisibleByThree && isDivisibleByFive) {
            return "FizzBuzz";
        } else if (isDivisibleByThree) {
            return "Fizz";
        } else if (isDivisibleByFive) {
            return "Buzz";
        } else {
            return Integer.toString(i);
        }
    }
}

You shouldn't declare your variables at the top of the method in Java, declare them as close to their usage as possible instead; this is not C. Ideally you avoid initializing a variable to a dummy value by joining the declaration and first usage. Don't increment and compare in a while loop condition, do one thing at a time.

Ben...

Canada3485 Posts

May 14 2015 19:11 GMT

#12620

On May 14 2015 20:14 Manit0u wrote:
Well, if you want to be an academic then I believe studying maths or philosophy would be more beneficial. CS should be pretty much purely technical (technical in learn how it's done in the real world sense) studies that prepare you for work in the industry.

There's technical/vocational schools for the purely technical side of things.

How university CS is right now makes sense. It's more in line with most other scientific university degree programs. You learn the underlying concepts and apply them in projects and assignments (just like how in physics or chemistry you learn things in lecture and apply them in labs). There seems to be a contingent of people who think that CS is purely software engineering and programming, and that is definitely not the case. Software engineering is just a small part of CS. You can go through an entire CS degree without ever stepping foot in a software engineering class (outside of the intro stuff you likely run into in second year). Likewise, there are CS classes where you don't program at all. From a CS perspective, programming is just a means of exploring and applying concepts. The end goal when programming during a CS degree is to show that you can apply the knowledge you learned, not produce some thing that you have to sell and maintain.

The whole "employers wanting people that know a lot of the theoretical concepts in CS but also all the latest and greatest tools" issue isn't with universities or technical schools, it's an issue with the employers. They want it all while not being willing to put up the resources to get it all. They want both someone who is both well versed from a theoretical perspective, but also somehow magically knows all of the particular technologies the company is using so they don't have to train them. They look to fresh graduates and expect them to know specific technologies that are not really something that are useful outside of a business environment.

Prev 1 629 630 631 632 633 1031 Next

Please or register to reply.

The Big Programming Thread - Page 631

Completed

Ongoing

Upcoming