How Python runs your code

Python is one of the most widely used programming languages in the world — for automation, web back ends, data analysis and machine learning. It is famous for being readable. This first lecture looks at what actually happens when you run it, because a clear mental model makes everything that follows easier.

Two ways to run Python

Install a recent version of Python 3, open a terminal and type python3 (on Windows, py). You are now in the interactive shell, also called the REPL — read, evaluate, print, loop. Type an expression and see the result immediately:

>>> 19.5 * 3
58.5
>>> "Euler" + "Skills"
'EulerSkills'

For real programs, write code in a file and run it as a script:

# greet.py
name = input("What is your name? ")
print(f"Hello, {name}! Welcome to Python.")

python3 greet.py

The shell is ideal for experiments; scripts are how you build tools you can rerun, share and test.

What happens when you press Enter

The standard implementation of Python, CPython, doesn’t run your text directly. It passes through a short pipeline. Scroll through it with a small program:

This is why Python is often called an interpreted language, even though there is a compile step: the bytecode is interpreted by the virtual machine rather than turned into native machine code.

You can watch the virtual machine work. Step through the real bytecode of two tiny programs, and notice that an if statement becomes nothing more than a conditional jump:

Two kinds of errors

Because parsing happens before execution, errors fall into two families.

Syntax errors are found while reading the code. Nothing runs at all:

print("Total:", total
# SyntaxError: '(' was never closed

Exceptions happen while the program is running, when an operation can’t be carried out — a name that doesn’t exist, adding text to a number, dividing by zero:

Traceback (most recent call last):
  File "report.py", line 12, in <module>
    summary = build_summary(orders)
  File "report.py", line 7, in build_summary
    return total / count
ZeroDivisionError: division by zero

Read a traceback from the bottom up. The last line says what went wrong (ZeroDivisionError: division by zero). The line above it says where (report.py, line 7). The lines above that show how the program got there — the chain of function calls.

Run five small scripts, each with one mistake, and compare what happens before the error:

NameError

A name was used before it was defined — often a typo.

TypeError

An operation was applied to the wrong type, such as "3" + 4.

ValueError

The type is right but the value isn’t, such as int("abc").

IndexError / KeyError

A list position or dictionary key doesn’t exist.

AttributeError

An object doesn’t have the method or field you asked for.

Is Python slow?

Executing bytecode one instruction at a time costs speed — pure-Python loops can be tens of times slower than equivalent code in C or Java. In practice this matters less than you might think:

• most business scripts spend their time waiting — for files, networks and databases — not calculating;
• performance-critical libraries such as NumPy, pandas and PyTorch are written in C, C++ or CUDA, and Python just orchestrates them;
• recent Python versions have made the interpreter significantly faster, and work continues.

Next, we look at the most fundamental idea in Python’s model of the world: names, objects and the difference between them.