pierd.github.io


Project maintained by pierd Hosted on GitHub Pages — Theme by mattgraham
27 August 2018

PyCon AU 2018

The company I work for was nice enough to provide me with an entry to PyCon AU 2018 conference. I’ve spent three days there (Aug 24-26).

Update for my grannies: people are nice, food is average, the weather was subpar but I’m having a good time.

Here are the talk that I’ve attended + some random notes.

PyCon AU organisers were nice enough to publish all videos on YouTube - you can find them here.


Friday August 24

Agloe - What the map makers of the 1930s can teach us about protecting our data in 2018

Canary tokens - when opened they trigger an alarm.

Few ideas where they can be placed:

Canary tokens don’t provide security by themselves but they have very good signal to noise ratio.

Privacy Preserving Record Linkage

First of all I like the history part on Brian’s website - I’m thinking about stealing the idea.

The talk itself - interesting problem I wasn’t aware of.

The main idea: splitting names (or other details) into N-grams and put them into bloom filter to anonymise the data but perserve the basic properties that allow for Record Linkage.

Libraries available on GitHub:

Security Architecture from Ancient Times

I highly recommend watching this one. Key takeaways:

Desirable control attributes:

Learning from the mistakes that even big projects make

Interesting but TBH I have not made any notes

Demystifying LoRaWAN with PyCom

Interesting but TBH I have not made any notes

Workplace Environment Sensing with Python

A really cool project. I liked the extra usage as a marker for interruption acceptance.

Automating Your Home with Python, Raspberry Pi and Homekit

pyhap + Raspberry Pi + AirCon + iPhone = MAGIC

Education Seminar Student Showcase

I watched only the first 2 presentations before heading to The Panel. From what I’ve seen it’s Machine Learning all the way.

  1. Playing noughts and crosses with the power of Machine Learning. Inspired by MENACE: the pile of matchboxes which can learn.
  2. Optimising flash cards style learning with Machine Learning (finding good spot for repetition ~ 40% recall).

The Panel

Interesting but TBH I have not made any notes


Saturday August 25

Remote First, Not Second Best

I wanted to listen to this talk since I worked remotely for few years.

“Culture eats methodology for breakfast”

Right tools are crucial. Synchronous communication is overrated and technology is not there yet anyway.

All in the Timing: Side-Channel Attacks

This one got me really interested. I’m going to try a timing attack in practice - stay tuned!

Few side-channel attacks mentioned in the talk:

Why you should care about types: How Python typing helped my team scale

Types are good. Types are pythonic.

Projects:

Duck typing? Protocols!

Type information for Standard Library? Typeshed!

Creepy, frivolous and beautiful art made with machines

Just watch it on YouTube!

Something to play with: https://github.com/google/deepdream


Sunday August 26

Write Perfect* Code First Time, Every Time

Visual Studio Code Python extension has breakpoints/stepping/debug console/etc.

It has real debug console (one can execute code in a running process). Internally it uses Python Tools for Visual Studio debug server.

Setup costs :(

Alternatives to VSC: pdb (plain old Python debugger), spyder (full blown Python IDE).

Reflections on the Creative Process - Illustrated with Watercolour Painting

The Creative Process:

  1. Planning - all the ideas how it’s going to come out
  2. Stress - is it really coming out as expected
  3. De-escalate - relax, think, be in a moment
  4. Decision moment - give up or keep going?
  5. Commit to the decision - life is short keep going
  6. Finishing touches -> stop when it’s good (don’t overdo)
  7. Allow appreciation time

Big life perspective: make a plan (prepare for bad events)

Key takeways:

FP demystified

Functional Programming in Python live in Jupyter Notebook!

Inspired by Professor Frisby.

Source code available here.

Colossal Cave Adventure in Python… in the browser!

Adventure is a really old (probably the first) text adventure game implemented in Fortran IV. Chris created a Fortran IV interpreter to run it and he hooked it up together behind Twillio API. At time of writing it was accessible via text message on +61 488 847 569 or +1 (669) 238-3683.

He also put it online in a Python interpreter in JavaScript (yes, Fortran IV in Python in JavaScript) - hilariously slow.


That’s all folks! See you next year…

tags: pycon - python