Evolution of Modern Python

Keynote, 2021-01-30

Alexander Hultnér, BelPy Online 2021

@ahultner

hultner.se/

Alexander Hultnér

Founder of Hultnér Technologies

linkedin.com/in/hultner

@ahultner

hultner.se/

Introduction

Focus forward on Modern Python's (3) evolution, not the transition from Python 2

Timeline

Python 3 release, December 3, 2008
Python 2.7 release, June 2009
Python 2 EoL, January 1, 2020

@ahultner

hultner.se/

Where it all started

2015-2016

Heliospectra

LED Grow Lights
Smart control systems for commercial greenhouses
Even worked with ESA (European Space Agency) in the Eden-ISS project
Anyone seen "The Martian"?

@ahultner

hultner.se/

@ahultner

hultner.se/

Where it all started

Next generation control systems

Old software

Desktop app in Python 2 aimed towards researchers.

New system

Adapted for commercial greenhouses.
Requires industrial scale, in-facility hardware, remote control, third party integration.

So can we use Modern Python, i.e. Python 3?

@ahultner

hultner.se/

Where it all started

Next generation control systems

New system

Modular architecture
Inter-process communication
enables multiple versions, and languages
Python 3.4, and then 3.5
Python 2 was never needed

@ahultner

hultner.se/

I'm fully convinced, Python 3 is great!

Python 3.4

Release Date: March 17, 2014

Significant features

PEP 435, a standardized "enum" module
PEP 3156, a new "asyncio" module, a new framework for asynchronous I/O
PEP 428, a "pathlib" module providing object-oriented filesystem paths

@ahultner

hultner.se/

And many more in the release notes

Python 3.5

Release Date: Sept. 13, 2015

Significant features

PEP 448, additional unpacking generalizations,
Splat operators: *[…], **{…}
PEP 492, coroutines with async and await syntax
PEP 484, the typing module, a new standard for type annotations
PEP 441, improved Python zip application support

@ahultner

hultner.se/

And many more in the release notes

Time to level up!

2017–2018

Prototyping new application
Great opportunity to utilise new features
- f-strings
- type-annotations
- dataclasses
Weekend project
- dataclasses with run-time type-checker and JSON parsing/formatting
GothPy
PyCon SE

@ahultner

hultner.se/

Python 3.6

@ahultner

hultner.se/

Release Date: Dec. 23, 2016

Significant features

PEP 498, Literal String Formatting, f-strings
PEP 468, Preserving Keyword Argument Order
PEP 530, Asynchronous Comprehensions
Ordered dicts (de-facto standard)
dataclasses ("backport" via pip install)
math.tau = 2π 😉

And many more in the release notes

Python 3.7

@ahultner

hultner.se/

Release Date: June 27, 2018

Significant features

Ordered dicts (official standard)
PEP 557, Data Classes (built-in)
Two-way iso format for date-times
Significant speedups

And many more in the release notes

A new dawn

@ahultner

hultner.se/

2019-present

Reincorporated my business
Hultnér Technologies AB

Python 3.8 Walrus Operator ":=" article
- More popular than I'd ever imagined
- Ranks 1-2 on Google for "walrus operator"
- Passed 100 000 views some time back
- Featured by medium curators, newsletters,
  Python Bytes Podcast, various news sites, etc.
Property Based Testing, Foss-North & PyCon SE

A new dawn

@ahultner

hultner.se/

2020, covid hits

Some good things

Python 2 EoL, hence this talk
Remote conferences increase accessibility, I've been able to talk at conferences every month.
(But I still miss the social aspect)
Python 3.9 is out
Lots of interesting things in the Python world like Schemathesis and Pyston2
Less time commuting => more time to experiment

Schemathesis

@ahultner

hultner.se/

A better way to handle API testing

Uses OpenAPI to automatically generate tests
Easy to use!
Based on randomised, property based testing
EuroPython Talk on YouTube
GitHub Repo, Docs
Test more with less!

Papero.io

@ahultner

hultner.se/

About experimenting more, with Magnus Junghard

Backend in Python 3.8 & 3.9
Async I/O, FastAPI Starlette, async task workers
Type annotations everywhere, mypy
- Automatically generated TypeScript API interfaces based on backend types & schemas
- Pydantic runtime type-checking
Randomised, property based testing
Deliver better results faster!

Papero.io

@ahultner

hultner.se/

About experimenting more, with Magnus Junghard

Okay, so what is Papero.io?

Create beautiful PDFs using dynamic HTML, CSS and JavaScript, or why not Django, maybe Flask?

Turn-key API as a service, or on premise.

Closed beta list opened recently.

Papero.io/beta for waitlist

Python 3.8

@ahultner

hultner.se/

Release Date: Oct. 14, 2019

First release with Steering Council

Significant features

PEP 572, Assignment expressions, "Walrus operator"
PEP 570, Positional-only arguments
Stabilize

More in the release notes

Python 3.9

@ahultner

hultner.se/

Release Date: Oct. 5, 2020

Significant features

PEP 602, Stable annual releases
PEP 584, Union Operators in dict
- Merge dicts with a|b, instead of {**a, **b}
PEP 585, Type Hinting Generics In Standard Collections
- Use "list" instead of "typing.List", etc
PEP 617, New PEG parser for CPython