Problem with "delete[]"

[melawrghk]

Hi everyone, I'm having some problems with deallocating memory on the heap. Here is a chunk of my code:

void DynArray::resize(int new_size){
  assert(new_size>=0);
  EType *new_array;
  new_array=new EType(new_size);
  int i=0;
  if (new_size<=sizeM){
	 while (i<new_size){
      new_array[i]=storageM[i];
      i++;
    }
	delete [] storageM;
        storageM=new EType[new_size];
    for(i=0; i<new_size; i++){
		storageM[i]=new_array[i];
	}
    delete [] new_array;
  }
  else if (new_size>sizeM){
	while (i<sizeM){
      new_array[i]=storageM[i];
	  i++;}
	delete[] storageM; //This line causes problems
	cout<<"1";
	storageM=new EType[new_size];
	for(i=0; i<new_size; i++){
		storageM[i]=new_array[i];
	}
	cout<<"2";
	sizeM=new_size;
	cout<<"3";
    delete[] new_array;
    }

}

The function is supposed to resize the array to the specified length and copy known elements into it. I don't really get why it's not working, it compiles fine, so I guess it's not my syntax, but I don't know.

Thanks in advance!

[Lindley]

new_array=new EType(new_size);

This does not construct an array of EType, it constructs a single EType using the constructor which takes an int argument.

By the way, is there a reason you aren't just using a std::vector<EType>?

[melawrghk]

Oh my goodness, I can't believe I did that. And overlooked it many times...

I'm not using a vector because the lab focuses just on dynamic arrays and we're restricted to using those.

Thanks a lot!

[Paul McKenzie]

Originally Posted by melawrghk

Oh my goodness, I can't believe I did that. And overlooked it many times...

I'm not using a vector because the lab focuses just on dynamic arrays and we're restricted to using those.

Why are you allocating twice? All you need is one allocation.

 new_array=new EType[new_size];
  int i=0;
  if (new_size<=sizeM){
	 while (i<new_size){
      new_array[i]=storageM[i];
      i++;
    }
    delete [] storageM;
   storageM = new_array;

// None of this stuff below is necessary

/*        storageM=new EType[new_size];
    for(i=0; i<new_size; i++){
		storageM[i]=new_array[i];
	}
    delete [] new_array;*/

Regards,

Paul McKenzie

[JohnW@Wessex]

Originally Posted by melawrghk

I'm not using a vector because the lab focuses just on dynamic arrays and we're restricted to using those.

But std::vector is a dynamic array

[superbonzo]

Originally Posted by JohnW@Wessex

But std::vector is a dynamic array

these teachers are very surprising... I mean, it's reasonable to teach how arrays work but why do they insist in asking to code std::vector like classes/templates ? maybe they think that the proper design of such a class/template is a newbie-affordable task ?

[Paul McKenzie]

Originally Posted by superbonzo

you're maybe they think that the proper design of such a class/template is a newbie-affordable task ?

The only reason that I can see is that the teachers are trying to weed out the potential good programmers from the weak ones, so that the weak ones drop the course. So they introduce (which IMO, doesn't need to be introduced) coding all sorts of things, asking the students to reinvent the wheel.

Anyway, the curriculum should be focusing on proper usage of the standard library, algorithms, etc. following somewhat in the way books such as Accelerated C++ approach learning the language. C++ has grown beyond the bounds of what the usual C++ curriculum looked like 12 or 15 years ago. You didn't have sort(), std::vector, reverse(), iterators, etc. back in those days, so back then, you had the traditional "write a string class..." and "write a function to reverse...", and "write a dynamic array..." assignments. Unfortunately, that very same curriculum has not changed in many teaching environments. The teachers and/or the ones responsible for the curriculum still believe that it's 1990.

My belief is that the curriculum needs to be changed to reflect what the C++ language means today. You can still have a good C++ beginner class that focuses on the fundamentals, and give good, meaningful assignments that stress usage of the containers and algorithms.

Regards,

Paul McKenzie

[JohnW@Wessex]

Originally Posted by Paul McKenzie

Anyway, the curriculum should be focusing on proper usage of the standard library, algorithms, etc.

I agree, I've always thought that the better way would be to get the student used to many different types of standard containers & algorithms, and then later on, for the more advanced students, posing the question "If you were to implement a vector/string/map style class, how would you go about it?"

The only advantage I can see to introducing home made containers early on is that when you get to use the STL you think, "Thank God I don't have to do all that new/delete/pointer code any more!".

[Paul McKenzie]

Originally Posted by JohnW@Wessex

The only advantage I can see to introducing home made containers early on is that when you get to use the STL you think, "Thank God I don't have to do all that new/delete/pointer code any more!".

Maybe, but the problem with this is that the "let's make them do this hard stuff first" type of teaching discourages many students, maybe turning them off from C++, and possibly programming in general.

Regards,

Paul McKenzie

[JohnW@Wessex]

Originally Posted by Paul McKenzie

... type of teaching discourages many students, maybe turning them off from C++,...

That's probably why all the C++ critics constantly go on about the lack of garbage collection. They bail out of C++ when it's being taught like C and think that's all there is to it.

(To any Java programmers reading this who are learning C++; Once you start coding in modern C++, garbage collection & manual memory management largely become non-issues)