As always, this year’s PyCon was full of interesting technologies and ideas. I took notes diligently and thought hey, why not just share them with all of you? Here’s a no-bullshit, no-filler list of all the actionable items that I noted down on my trip; in chronological order (and not order of importance).
- It would be really cool to have a site that serves as a public collection of A/B test experiments and results from various companies. Someone should make this!
- glom is a Python library for querying nested data structures (meaning dicts in lists in dicts in dicts in […]). It can replace tens of lines of data mangling and error handling with a one-line query. It’s also serious enough to have a built-in debugger for its query strings!
- Check Quilt for version-controlling datasets.
- We should reconsider whether we want to keep any secrets in environment variables because they leak way too easily. Instead, we should use secret management provided by Rancher/Kubernetes to get Vault tokens to our apps.
- Check PlusPlus for managing knowledge sharing (workshops and such) in your company.
- It’s time for us to adopt natural language-based feature tests. Same goes for all teams with enough non-technical stakeholders. The best tool for this seems to be pytest-bdd.
- Pyre is a better performing but less fully-featured alternative to mypy, the type-checker.
- We should set up a dedicated space in the office for pair programming where people can work without interruption or the fear of disturbing others — with a monitor, two keyboards and two mice.
- Confir collects and keeps structured data about conferences around the world.
- 4degrees.ai might be a nice tool for managing professional contacts. It scans people’s blogs, Twitter, and stuff to suggest topics to use as an excuse to talk to them.
- code-or-die is a cool idea for a ‘board game’ you interact with via a JSON API.
- I loved James P’s slide design for the code-or-die talk. I want to use this little design element of keeping half the text in place through slide transitions.
- We should keep checklists for people to go through before they release new projects to production. This should probably be part of some sort of human service registry (such as The Zoo, which is our soon-to-be open-sourced solution to this at Kiwi.com — stay tuned for news on this project!)
- DataSine personalizes marketing communications based on machine learning magic. It guesses each customer’s persona, what kind of tone they prefer, and so on.
- Woven Teams are consultants for perfecting hiring processes. For instance, they help with designing technical interview tasks.
- Datmo is version control for neural nets.
- The Python Developers Survey of 2017 is well worth a read.
- Hynek Schlawack has a library for configuring apps via environment variables called environ. It seems like the best one by far for this purpose.
- The rows library by Álvaro Justen looks great for converting tabular data between different formats, both in Python code and as a CLI tool.
- Read how.complexsystems.fail on, well, how complex systems fail.
…and that’s it! I encourage any and all PyCon goers reading this to leave a response with their own notes, even if it’s just a collection of links, and even if it’s just one item. We, both as a community and an industry, would certainly benefit from more simple bits of information from conferences being available for all online.
Shameless plug: I myself publish notes from some conferences I go to on GitHub which I think is a great way to go about this.