new vs. malloc

[Asapi]

Compared to malloc(), why would one say that the 'new' operator is type-safe
in a book? Thanks!

 

[Jeffrey Schwab]

Compared to malloc(), why would one say that the 'new' operator is type-safe
in a book? Thanks!

1. malloc() returns void*, but new returns the proper type.

2. malloc'd objects must be manually {con,de}structed.

 

[tom_usenet]

Compared to malloc(), why would one say that the 'new' operator is type-safe
in a book? Thanks!

malloc returns void*, new returns a pointer of the type of the object
created. e.g.

int* ip = malloc(sizeof(short));
//whoops! no compile time error, but runtime one

vs
int* ip = new short;
//compile time error, short* not convertible to int*

Tom

C++ FAQ: http://www.parashift.com/c++-faq-lite/
C FAQ: http://www.eskimo.com/~scs/C-faq/top.html

 

[Adam Fineman]

Compared to malloc(), why would one say that the 'new' operator is type-safe
in a book? Thanks!

"Type-safe" loosely means that if you use the wrong type, the compiler
can tell you about it. Look at the following examples:

int* ip1 = static_cast<int*>(malloc(10 * sizeof(char)));

int* ip2 = static_cast<int*>(new char[10]);

The second line will fail to compile, whereas the first one is just fine
(at least according to the compiler).

- Adam

 

[tom_usenet]

malloc returns void*, new returns a pointer of the type of the object
created. e.g.

int* ip = malloc(sizeof(short));
//whoops! no compile time error, but runtime one

Note that the above is C code (where malloc is idiomatic), the C++
would be:

int* ip = static_cast<int*>(malloc(sizeof(short)));
//whoops! no compile time error, but runtime one

which is even more painful to write.

Tom

C++ FAQ: http://www.parashift.com/c++-faq-lite/
C FAQ: http://www.eskimo.com/~scs/C-faq/top.html