About dynamic allocation of memory

Suppose I was working for some search engine like Google where every every line of code needs to be at its finest.

Then if I were writing code for such an engine, would I use pointers to do everything instead of using regular variables? Because variables cannot be de-allocated whereas pointers can be de-allocated right? :confused:

What's a brief estimate for how many variables you can allocate? Just curious.

This is C++; you're going to be using the standard containers.

Quote:

---

Originally Posted by tabstop

This is C++; you're going to be using the standard containers.

---

Then where are pointers used other than to pass by value in functions (and even in that pass by reference is more often used)?

To implement the containers and smart pointers, sometimes you might even implement using them because the standard ones don't meet your requirements, whether directly or as components for your own containers and other (template) classes.

That said, you might use a pointer for an optional argument, though there are alternatives too.

EDIT:
I suggest that you read Stroustrup's answer to the FAQ: How do I deal with memory leaks? It is a bit dated in that it mentions the obsolete auto_ptr that has been superseded by unique_ptr, but it still holds true.

Okay so pointers can be useful. But is it not worth it trying to use dynamically allocated variables? Like if you were working with a lot of data (you have to get rid of the variables at some point right)? Or can you get rid of variables by using

Code:

---

{ }

---

(block) so that out of the block the variables lose their scope and get de-allocated? I know this is not necessary but suppose, just suppose that if we were making code for some really small machine or for something really important like a search engine? I'm just curious. So how would programmers deal with having very concise memory?

Heh the forum saw my braces and thought I was trying to post code but that's cool, more forums should have this.

You use dynamic allocation when either of these is true:
- You don't know at the point the program is started, just how many things you need.
- You know that you won't want to keep the data for the lifetime of the program.

So for example, an image editor would use dynamic allocation on both counts because:
- it doesn't know how big each image is going to be.
- you won't want to keep the old image lying around when you edit the new image.

Yes, putting your variables in as small a scope as feasible encourages both readability and the deallocation of memory once it is no longer needed. Besides, you can always write the component to manually control deallocation if you really need to do so. Consider std::vector, which has a shrink_to_fit member function, and if that doesn't satisfy you as it is not guaranteed to actually do the reallocation/deallocation, you can swap the vector with an empty vector that will shortly go out of scope.

Okay guys thanks