使用C++的疑惑

alarum

我的C++和数据结构是自学的，一直以来书上怎么写我就怎么做，到了论坛里发现大家贴出的源代码有一些习惯我不理解。比如说很多人用C++时从不用new和delete，而是一定要用malloc/free。这是为什么？我觉得malloc用起来远没有new方便。这只是一些人的习惯还是有什么原因？另外我的数据结构学的是——殷人昆（用面向对象方法描述）朋友建议我学一下严魏敏的，我看了一下，使用C语言描述的。有必要重学一遍吗？

现在在家，没地问问题。想听听大家的意见，谢谢

lenovo

你在哪里看的“很多人用C++时从不用new和delete，而是一定要用malloc/free”？
我很想看一下。

tp801

Most of the C++ developer *will* like to use new/delete operator for memeory management, few of them *may* use malloc/free(or other third party memory management library) for some advanced system programming, but this is rarely happened.(I assume this kind of users are really C++/system expert and intend to do that in this way).  Other than that, you can/should follow the normal way(new and delete) to do your C++ programming, especially you are new to C++.

Hope this help.

alarum

说有很多人可能是过了一些，不过确实有一些人这样写。可能是因为我几乎只用new/delete，所以读到下面这段代码会比较敏感。

1. 
   
2. Status InitList_Sq(SqList &L)
   
3. {
   
4. L.elem=(elemtype *)malloc(LIST_INIT_SIZE*sizeof(elemtype));
   
5. if(!L.elem) exit(OVERFLOW);
   
6. L.length=0;
   
7. L.listsize=LIST_INIT_SIZE;
   
8. return OK;
   
9. }
   

复制代码

这是一位仁兄的代码，他用的是C++。（只是引用一下，没别的意思）我是个完全自学的新手，暂时身边也找不到明白人，所以疑问也比较多。在这里只是想确认一下我的想法
[/quote]

soloarrow

我个人的理解是：那段话的含义是针对“for memeory management”，对于内存上的操作，高手们的确是自己开辟内存，再自己进行初始化的；而new在C＋＋中还有其他的用途。比如你举的那段代码，可以自己进行控制“if(!L.elem) exit(OVERFLOW); ”，而如果用new的话，要用异常来进行判断。
  还有就是new/delete操作是通用的c++编程规则，应该遵守，这个观点英文的最后一句也提及了。

我不懂C++

很多人使用malloc/free是因为觉得自己处理的是scalar，这个也无所谓。

tp801

[quote]原帖由 "alarum"][/quote 发表：


[quote]原帖由 "alarum"][/quote 发表：



OK, let's go a little bit deeper for the "new/delete" operator. As you might know, "new" operation is actually calling "malloc" to do the underlying implementation, while it also maintains a 'header' to   keep track of the memory usage, this is good for memory management, but sometimes(in a rare situation), it's a overhead, especially when you are designing a real-time/system program( I think the code you attached here is actually for such purpose, otherwise, it's not a good C++ programe ). Other than that, "new" and "malloc" are almost the same(of course, "new" will trigger a constructor of the allocated object, whilc "malloc" will not, but this is out of this topic here).

One more time, as a regular C++ programer, you are encouraged to use "new/delete" insteadof "malloc/free" unless you had to do so.

Further more, I want to clarify sth. about data structure and C++/C. I don't think they have direct relation to each other, C++ or C are just used as a language to help to descript the data structure, but they are NOT part of the data structure, you can use any other language you are familiar with to do that also. Peasonally speaking, I like to use C to learn data structure.

Hope this help.

我不懂C++

you are wrong, to use operator new will not add any overhead, only operator new[] will have some overhead, however, for PODs, some compilers will optimize off these overhead, and for non-PODs, you cannot avoid calling the dtor when destroying.
You may have some single step in your IDE to check what I say

whyglinux

在C++程序中，malloc()/free()的作用有限，只适用于对于基本数据和简单的结构等类型（PODs）的内存分配和回收，一般不适用于类。因为类一般有构造函数或析构函数，在创建或者删除类对象的时候需要自动被执行。所以说楼主说的“很多人用C++时从不用new和delete，而是一定要用malloc/free”确实是言过其实，除非你在C++中不使用类。比如楼主提供的代码中如果 elemtype 是一个类而不是一个POD的话，不管出于什么理由，不用new/delete而用malloc()/free()来分配和释放内存都是说不过去的。

对于PODs的内存分配和释放，出于与C代码的兼容考虑，可以使用malloc()/free()。

我不懂C++

不过有一个原因，使得malloc/free还有市场：内存分配不出的时候，malloc返回NULL，而new理论上应该是抛出异常。在某些兄弟眼里，异常可是洪水猛兽。