The Big Programming Thread - Page 37

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.

EscPlan9

United States2777 Posts

February 13 2011 22:27 GMT

#721

JGrasp is incredibly simplistic and lacks a lot of convenience functionality. However, it is the easiest to use for basic applications. However, I do not recommend it for anything GUI based - it requires manually defining every portion.

I never tried out Eclipse, but when I worked with my group creating a multi-threaded network-based blackjack application, one guy created the GUI straight from Eclipse. Or it may have been JavaBeans? Anyways, definitely not with JGrasp. (I coded the logic and deck handling/manipulation portion of the program using JGrasp though.)

Disregard

China10252 Posts

February 13 2011 22:35 GMT

#722

Eclipse by far, though the interface takes some getting used to but it has great functionality.

jsmith

United States18 Posts

February 14 2011 00:21 GMT

#723

Ya right now I am not doing anything GUI based, just some basic console applications.

Kambing

United States1176 Posts

February 14 2011 01:46 GMT

#724

On February 14 2011 06:07 jsmith wrote:
you guys got any suggestions for a good ide to do java in? is it pretty much just use eclipse?

Yup as mentioned, Eclipse is the de facto standard for "real" Java development. If you program in Java, you'll eventually want to learn how to use eclipse.

If you're starting out with programming or just want to streamline your learning before moving onto Eclipse, then a lightweight IDE like the aforementioned JGrasp or an educational IDE like DrJava (my recommendation) or BlueJ would also be appropriate. DrJava is particularly nice because of the interactive window that lets you quickly experiment with Java language constructs.

Blisse

Canada3710 Posts

February 14 2011 02:01 GMT

#725

Does anyone have a recommendation for an IDE for Python? I currently use Komodo, but it's very weird.

And anyone have any advice for the Canadian Computing Competition? I have to write it on March 1. And I have to relearn Python by then. I don't expect to make it to the Second Stage, where you must use Java and C, so Python is fine.

wO-ZiGouNeT

United Kingdom21 Posts

February 14 2011 09:31 GMT

#726

http://www.wingware.com/ is pretty decent, I've only used the free version but the debugger is pretty much all I ask for from an IDE.

snpnx

Germany454 Posts

February 14 2011 09:36 GMT

#727

On February 14 2011 06:07 jsmith wrote:
you guys got any suggestions for a good ide to do java in? is it pretty much just use eclipse?

Yeah, it's pretty much Eclipse.
Another good IDE which I wonder why it wasn't mentioned is NetBeans.

Both are good, Eclipse is a bit better in my eyes, but Netbeans might be a bit easier if you just start with java.

Blisse

Canada3710 Posts

February 14 2011 22:04 GMT

#728

On February 14 2011 18:31 wO-ZiGouNeT wrote:
http://www.wingware.com/ is pretty decent, I've only used the free version but the debugger is pretty much all I ask for from an IDE.

Ugh. Not Wing again.

I went to Komodo to get away from Wing.

wO-ZiGouNeT

United Kingdom21 Posts

February 14 2011 23:09 GMT

#729

Notepad++ / emacs to edit the source, wing to debug xD Haven't used it since uni tho, I don't get to enjoy working with python : /

jaydubz

21 Posts

February 16 2011 00:28 GMT

#730

So I've returned to C++ after first learning it 3 years ago to work on a personal project and I've ran into a problem.

I've created three classes. For simplicity we can call them class a, class b, and class c. The problem is each one contains the other (causing a circularity issue, from what I understand).

i.e.:

class a contains a private member: c obj

class b contains a private member: a obj

class c contains a private member: b obj

Currently, using Dev C++ and trying to compile this won't work. This is because each class contains a class that hasn't been defined yet (from what I understand). Need someone to dumb down how to go about solving this problem. I can supply the header files if needed.

fabiano

Brazil4644 Posts

February 16 2011 00:36 GMT

#731

It looks like a problem in the design of your solution.

I think having the headers would help us to help you.

Chaosu

Poland404 Posts

February 16 2011 00:42 GMT

#732

im sure that putting:

class c; //this initiates the class for the compiler

class b;

start defining class a then b then c

will get you rid of some compiling errors but im not sure if this will solve problem entirely. of course you have to friend classes for access to their privates

jaydubz

21 Posts

February 16 2011 01:01 GMT

#733

Found source tags.

+ Show Spoiler +

#ifndef exit_h
#define exit_h

#include "location.h"
#include "exitList.h"
#include <string>
using namespace std;

class exit
{
	private:
		string direction;
		location leadsTo;
        public:
                //public stuff
};
#endif

+ Show Spoiler +

#ifndef exitList_h
#define exitList_h

#include "exit.h"
#include "location.h

class exitList{
	struct node{
		exit data;
		node link;
	} *p;
       public:
                //some public stuff
};
#endif

+ Show Spoiler +

#ifndef location_h
#define location_h

#include <string>
using namespace std;

#include "exit.h"
#include "exitList.h"

class location
{
	private:
		string title;
		string description;
		exitList * exits;
	public:
               //some public stuff
};
#endif

As for friend, adding setter/getter functions for all private members would eliminate this need I believe?

Chaosu

Poland404 Posts

February 16 2011 01:36 GMT

#734

well i put that code in devc++ and merged those headers into one file and friended exit and exitList to location i only got those errors:

in class exitList:
'exit' does not name a type
field 'link' has incomplete type

+ Show Spoiler +

#ifndef combined_h
#define combined_h

#include <string>

using namespace std;

/*class exitList;
class exit;
class location;*/

class location
{
friend class exit;
friend class exitList;
private:
string title;
string description;
exitList * exits;
public:
//some public stuff
};

class exit
{
private:
string direction;
location leadsTo;
public:
//public stuff
};

class exitList{
struct node{
exit data;
node link;
} *p;
public:
//some public stuff
};

#endif

so your "circularity" issue is fixed, but you need to declare your classes and structures a bit better, not going to help you now because its 230 AM and im going sleep

jaydubz

21 Posts

February 16 2011 01:54 GMT

#735

Well, the error

'exit' does not name a type

is the error I'm currently receiving, among others (most likely because of exit class not compiling).

It seems combining them still maintains the issue: location contains an exit list, which contains an exit, which contains a location, etc.

If I declare

class exit; class exitList;

prior to declaring

class location{ }

the order of the errors change but the error still persists.

fabiano

Brazil4644 Posts

February 16 2011 02:47 GMT

#736

Ok, I figured out your problem I guess.

Its a design problem. You have the class Exit which I believe represents an exit of a location.

The thing is, the attribute leadsTo of the class Exit is highly dependent on what location are you currently in. This way leadsTo actually would have to have 2 values at the same time.

For example, if there are Location A and B, and an Exit E which links both locations, the value of E.leadsTo depend where you are currently. If you are in A, leadsTo point to B, if you are in B, E.leadsTo points to A.

You could solve it by having 2 attribute leadsTo in the Exit class, however it wont solve your design problem as a whole.

The other problem is that you wrote 2 unnecessary classes: Exit and ExitList.

A Location is nothing more than a room with its attributes (title, description) and a list to other locations (the exits). So a class describing an exit is redundant and caused you the crossed references.

It is hard for me to explain it with text, and even harder when I try to put my thoughts in english, so I hope you can understand me by just looking at the classes I wrote below, which I believe should solve your problem from a design standpoint.

+ Show Spoiler [LocationData.h] +


#ifndef ___LOCATIONDATA_H___
#define ___LOCATIONDATA_H___

class LocationData {
    public:
    LocationData();

    private:
    std::string mTitle;
    std::string mDescription;
};

#endif

+ Show Spoiler [LocationList.h] +


#ifndef ___LOCATIONLIST_H___
#define ___LOCATIONLIST_H___

#include "LocationData.h"

class LocationList {
    public:
    LocationList();

    private:
    LocationData mData;
    LocationList mNext;
};

#endif

Hope it helps!

Edit: thinking again, I think what I did does not solve your problem at all

, but I won't delete the post since it could give you a hint or something.

djcube

United States985 Posts

February 16 2011 06:56 GMT

#737

EDIT- sorry for the clutter! I've thought up a possible solution.

jaydubz

21 Posts

February 16 2011 11:48 GMT

#738

The exit list class is just a linked list I designed. I did this because the std list I was originally trying to use would not play nicely when trying to append. I probably needed to overload a few operators to get it working but I felt it would be better practice to create my own data structure.

Frigo

Hungary1023 Posts

February 16 2011 14:02 GMT

#739

On February 16 2011 09:28 jaydubz wrote:
I've created three classes. For simplicity we can call them class a, class b, and class c. The problem is each one contains the other (causing a circularity issue, from what I understand).

Predeclare your classes and use member pointers instead of member variables. Just think about it, what is the size of class A if it recursively contains itself?

Split each of your (non-template) classes to a header file (.h or .hpp) and an implementation file (.cpp). Predeclare the class in the header file before you include anything to guarantee all required types are known, and are at least incomplete types. Use member pointers or some smart pointer like shared_ptr instead of member variables then. You can't use incomplete types to instantiate, as member variables or calling their functions, and in a lot of cases, as template arguments. However you can use them as pointers and member pointers. This makes circular references between classes possible. In fact, this is the best way. It is possible to solve the problem with template trickery, but it is MUCH more difficult and messy.

You can include only headers and compile cpp files separately then link them together (the nice way), or just include cpp files after header files (mostly used for template classes, in case they are split at all).

+ Show Spoiler [A.h] +


#ifndef A_h
#define A_h

//
//	predeclarations
//

class A;

//
//	includes
//

#include "C.h"

//
//	class A
//

class A
{
	public:
		A ();
		virtual ~A ();
		double someFunction () const;
		[...]
	protected:
		C* c;
};

#endif

+ Show Spoiler [A.cpp] +


#ifndef A_cpp
#define A_cpp

#include "A.h"

A::A () :
	c (new C())
{
	[...]
}

A::~A ()
{
	delete c;
}

double A::someFunction () const
{
	c->doSomething();
}

#endif

+ Show Spoiler [C.h] +


#ifndef C_h
#define C_h

//
//	predeclarations
//

class C;

//
//	includes
//

#include "B.h"

//
//	class C
//

class C
{
	public:
		C ();
		virtual ~C ();
		double someFunction () const;
		[...]
	protected:
		B* b;
};

#endif

+ Show Spoiler [C.cpp] +


#ifndef C_cpp
#define C_cpp

#include "C.h"

C::C () :
	b (new B())
{
	[...]
}

C::~C ()
{
	delete b;
}

double C::someFunction () const
{
	b->doSomethingOrAnother();
}

#endif

B similarly...

dvide

United Kingdom287 Posts

February 16 2011 15:20 GMT

#740

First thing, you don't need to include exitList.h in exit.h. You never mention exitList directly in that file. In exitList.h you don't need to include location.h. In location.h you don't need to include exit.h.

Now, in location.h you're including exitList.h which itself includes exit.h (ignoring the fact that you're including exit.h directly anyway). If you follow the logic of that, in location.h you're using the class location before it is even defined (by including exit.h, which uses location). What you should do is remove the include to exitList.h here and instead replace it with a forward declaration of the class (i.e. class exitList;). Where you include exitList.h is in location.cpp instead.

The reason you can break the circular dependency here and not elsewhere is because you're actually using exitList*, not an exitList itself. So this means any translation unit that includes location.h doesn't actually need to know the full definition of the class exitList (except for location.cpp, because here you'll actually create the exitLists). It only needs to know that there exists a class by that name, which is what you achieve with the forward declaration. By using the forward declaration, you're promising that there exists a definition of the class in another translation unit to be linked in. If you fail to provide it somewhere else you'll get a linker error instead of a compile error.

I hope that makes sense as to why you're getting this problem.

Prev 1 35 36 37 38 39 1032 Next

Please or register to reply.

The Big Programming Thread - Page 37

Completed

Ongoing

Upcoming