Resources

Stack

Properties

  • Small.
  • Fast.
  • Automatic.
  • Data is ordered: LIFO (Last In First Out).
  • Fixed size (1-8 MB).
  • Data is popped (removed) from the stack when function returns.
  • Allocation sizes must be known in advance.

Stack Overflow

Stack Overflow happens when:

  • Depth Problem
    • when creating too many frames in the Stack until the stack limit.
    • e.g. recursion with no stop condition.
    • fix: always have a base case.
  • Size Problem
    • when creating a frame too large.
    • e.g. try to create a huge variable.
    • fix: put large data in the Heap.

Heap

Properties

  • Large pool of memory (can grow to use most of available RAM).
  • Flexible.
  • Managed.
  • Data is not ordered: you program request space, the system finds a free block and returns the address.
  • Access data using its address which is tiny and lives on the stack.
  • Use cases:
    • when don't know how much space is need until runtime.
    • when data needs to survive after the function.
    • when data is too large for the Stack.

Memory Leak

How does the Garbage Collector work:

  1. Collector starts from the Stack and traces every reference chain.
  2. Anything reachable in the Heap is kept. Anything unreachable is deleted.

Problem: if the program keeps references to object it no longer needs, the Heap memory is no cleared because still reachable

Fix: Explicit cleanup. The garbage collector claim unreachable memory, making memory unreachable is the developer job.

Programming Language Variables Types

Primitive

Primitives / Pass by Values (numbers, booleans, ...)

  • When a primitive variable values is passed to another variable, the value is copied.
  • Now there are two separates variables with the same value in the Stack, modify one variable value will not impact the other one.

Objects

Objects / Pass by Reference (lists, dicts, classes, ...)

  • When an object variable value is passed to another variables, the reference is copied.
  • Now the two variables old the same reference (address) pointing to the same data in the Heap, modify one variable value and the other variable will be impacted.

Example in Python

Assignment statements in Python do not copy objects, they create bindings between a target and an object. For collections that are mutable or contain mutable items, a copy is sometimes needed so one can change one copy without changing the other.

Source: Shallow and deep copy operations - Python

# List of dicts
my_list = [
	{ "player": 1, "class": "knight" },
	{ "player": 2, "class": "pirate" }
]

# Normal Copy
# Assign reference of the list that contains references to the dicts
my_list_2 = my_list

# Swallow Copy
# Create a new list that contains the same references to the dicts
my_list_3 = copy.copy(my_list)

# Deep Copy
# Create a new list and new dicts
my_list_4 = copy.deepcopy(my_list)