Memory Allocation

[peridian]

Hello,

I may be remembering incorrectly, but I'm sure that in C++ you could declare two variables of the same name where the second declaration is within it's own code block. E.g.

void myfunction()
{
  int x = 5;
 
  //some code
 
  {
    int x = 3;
 
    //print x
  }
 
  //print x
}
 

In this case, the first time x is printed, it will refer to the x declared within the inner block, and the second time it will refer to the x in the outer block.

If this is correct, my question refers to the compiler and stack allocation.

I found out the other day that Java allocates memory for all local variables within a function block, regardless of whether the code actually reaches those declarations. Apparently, this is to make the compiler/runtime easier to code and manage. I'm assuming this is the reason you can not have a situation similar to the above code example in Java, as you would have two local variables by the same reference being allocated to the stack at the same time.

I wondered if this was the way all languages handled memory allocation. If this is the case in C++, my question is how does C++ allow two variables to be allocated to the same named reference within a function block? Does C++ allocate to the stack each time it enters any kind of code block? Or do pseudo/mask references get used, whereby the first x is internally considered as x1 and the second as x2?

Or something...

I have not a lot of knowledge on this area of coding, but it's something I've recently started wondering about for the purposes of memory optimization in my coding.

Regards,
Rob.

[weaknessforcats]

In C++, the operative word is stack.

That means as variables are encountered they are added to the stack. Only in C are all of the variables created on the stack at the beginning ofr the function.


C has a very simple stack frame where these variables are placed. The C++ version allows you do add to the stack frame as the function progresses. Exactly how this is done is left to the compiler writer. It could be your C++ compiler works just like your Java example but it is not required to do so.

[peridian]

Ahh, cool.

I had always done my coding on the understanding of memory being allocated to the stack as the declarations were encountered, so you can imagine my surprise when I found out about Java.

Interesting that C does it the same way as well, that's a useful tip.

Would it be fair to say that the majority of compilers do it as we have said for C++, but that it is dependant upon the compiler implementation? I.e. some compilers may do it the same as C and Java, but most don't.

Regards,
Rob.

[weaknessforcats]

I have no idea what the common stack implementation is.

As a programmer, I just follow the scoping rules and let the compiler generate the proper code.

Also, I don't believe Java has a problem with two variables of the same name so long as they are in different scopes.

[peridian]

Fair enough.

With regard to Java, it definitely does (or at least in 1.6) have a problem with duplicate variable names within the same function block, even if in separate scope. The below is an example of the compiler message you get:

Exception in thread "main" java.lang.Error : Unresolved compilation problem:
Duplicate local variable results


Thanks for the info.

Rob.

[weaknessforcats]

If that's reaaly true, then that is one big hole in Java. Big enough to prevent me from ever considering it for a real application.