C++ Masterclass & Complete Course Documentation

g++ -std=c++20 -Wall -Wextra -Wpedantic -Wconversion -fsanitize=address,undefined -O2 main.cpp -o app

Enables standard C++20, all warnings, strict pedantic checks, address & undefined behavior sanitizers, and optimization level 2.

Container	Random Access	Insert / Erase (Back)	Insert / Erase (Front/Middle)	Underlying Structure
`std::vector`	O(1)	O(1) amortized	O(N)	Contiguous Heap Array
`std::deque`	O(1)	O(1)	O(1) front/back, O(N) middle	Chunked memory blocks
`std::list`	O(N)	O(1)	O(1) if iterator known	Doubly-linked list
`std::map / std::set`	O(log N) search	O(log N)	O(log N)	Red-Black Tree (Ordered)
`std::unordered_map`	O(1) avg search	O(1) avg	O(1) avg	Hash Table

Prefer Stack over Heap: Stack allocations are instant, cache-friendly, and automatically freed.
Rule of Zero / Five: If you define a custom destructor, copy/move constructor or assignment, define all five (or prefer smart pointers so you need none).
Smart Pointers: Use std::unique_ptr by default. Only use std::shared_ptr when ownership is genuinely shared.
Pass by Const Reference: Pass non-primitive objects as const Type& or std::string_view to eliminate copies.

Prefer std::jthread over std::thread: Auto-joins on destruction and cleanly handles cooperative cancellation via std::stop_token.
Deadlock Prevention: Always use std::scoped_lock(mtxA, mtxB) when locking multiple mutexes simultaneously.
Spurious Wakeups: Never call cv.wait(lock) without a predicate lambda loop: cv.wait(lock, [&]{ return !queue.empty(); });.
Eliminate False Sharing: Align hot atomic counters to CPU cache lines using alignas(64).
Prefer Tasks over Threads: Use std::async(std::launch::async, ...) and std::future for concurrent calculations with automatic exception propagation.