Talk:Criticism of C++/Archive 1

Archive 1

list of problems with c++

Here's some problems with C++ that I have encountered. There's a few things that the language doesn't let you do, for no obvious reason. There's also a few things which it lets you do, that really it shouldn't, because they promote bad code and have no great advantage over doing things a different way. Finally, there are some things that just plain work in a weird way, which form a minefield for anyone attempting to use them who lack the experience to steer clear.

Missing features

• initialising non-integer static const variables inside a class header
• recover the size of an array at run time

Features that don't work (or don't work as you'd expect)

• you are allowed to override operators (&&), (||) and (,) but it is a really bad idea
• you can create your own type conversion operator()'s, but it is not good practice, as unintended results are likely
• you can call virtual functions from constructors or destructors, but they do not work as you would expect.
• the order of initialisation of global objects is undefined, which may cause errors if there are dependencies between them
• exception specifications generally aren't implemented, and in any case do not work the way you'd want them to
• redefining the default parameter values for inherited virtual functions doesn't work properly (they ain't virtual)
• single parameter constructors are used as implicit type conversion operators by default, which seems wrong
• you can use arrays of objects polymorphicly, but it doesn't work (deletion and indexing can both fail).

The use of templates to achieve certain affects could be seen as a hack to make up for some of the inadequacies of the c++ preprocessor.

213.120.94.99 (talk) 13:03, 31 July 2008 (UTC)

Low article quality

This article is not well recherched and requires serious rework before merged. 85.88.24.15 (talk)

The topic "Criticism of C++" is a requested article on wiki. Hence an effort has to be made to start somewhere so that it can be revamped by experts once the work gets started. Aandu (talk) 08:38, 26 December 2007 (UTC)

The following Topics need to be elaborated upon:

  No Dynamic type safety  
  Criticism of features that break Object Orientation
  Standardization drive 
  Maintenance
  Economy of expression
  Tools for mitigating issues with C.... Aandu (talk) 08:46, 26 December 2007 (UTC)

Let me add a topic for elaboration: In my experience, compared to C, C++ programs tend to end up vastly more bloated, overly-complex, and hard to follow. This seems to be from a proclivity towards major over-and-mis-use of Object Oriented programming as a universal panacea, where every stupid little thing becomes an elaborate class. Discuss. —Preceding unsigned comment added by 24.18.201.182 (talk) 11:43, 29 December 2007 (UTC)

While the article is not of high quality, I can certify there is no original research on it, thus I'm removing the warning about this. I'm an experienced C++ user, and I am expert enough to certify this (because the content of the article is commonly known among C++ users).--Blaisorblade (talk) 01:43, 20 June 2008 (UTC)

Btw, C++#Criticism seems to have better style than this article.--Blaisorblade (talk) 02:04, 20 June 2008 (UTC)

syntax for range

There is really syntax for range.

example:

#include<cstdio>

unsigned int n;

int main(){

 scanf("%u",&n);

 switch(n){

  case 0:

   puts("This is zero");

   break;

  case 1 ... 9:

   puts("It has one digit");

   break;

  case 10 ... 99:

   puts("It has two digits");

   break;

  default:

   puts("It has more than three digits");

 }

 return 0;

}

The above is not valid C++. I think the ranges in a case statement (1...99) is an extension of the GNU/GCC compiler. 195.14.207.109 (talk) 15:33, 3 June 2008 (UTC)

I confirm this, see [1]. --Blaisorblade (talk) 01:51, 20 June 2008 (UTC)

function as parameter

This is an example using a function as a parameter

#include<cstdio>

void swap(int&a,int&b){
	int c=a;
	a=b;
	b=c;
}

void sort(int*arr,int n,bool cmp(int,int)){
	for(int i=0;i<n-1;i++)
		for(int j=i;j>=0;j--)
			if(cmp(arr[j+1],arr[j]))
				swap(arr[j],arr[j+1]);
}

bool reverse(int a,int b){
	return a>b;
}

int main(){
	int list[]={1,2,3,4,5};
	sort(list,5,reverse);
	for(int i=0;i<5;i++)
		printf("%d\n",list[i]);
	return 0;
}

The 3rd parameter of sort(int*,int,bool(int,int)) is really a function. —Preceding unsigned comment added by Michaeldadmum (talk • contribs) 13:00, 12 January 2008 (UTC)

In the above sort function, "bool cmp(int,int)" is invalid. It has to be a function pointer. 195.14.207.109 (talk) 15:33, 3 June 2008 (UTC)

The correct way to define a function parameter, in accordance to C How To Program Second Edition by H. M. Deitel and P. J. Deitel, is:

<TYPE> (*<NAME>)(<PARAMS>)

So, the correct way to define a function as a parameter in the above function sort is:

void sort(int *arr, int n, int (*cmp)(int &, int &))

This is explicitly mentioned in the last section of chapter 7: "Pointers". Graham (talk, contrib, SIGN HERE!!!) 16:10, 20 July 2008 (UTC)