Fred Hebert

Property-Based Testing is one of the most powerful testing methods out there, and the Erlang ecosystem has some of the best frameworks available across all languages. Property-based testing has the potential to drastically increase the quality of software we ship by testing for issues we couldn’t even imagine.
Despite great tooling and promises, adoption is still somewhat limited. Simple examples are easy to write, but more complex ones can often feel like black magic. 

In this Q&A session, Fred Hebert invites John Hughes, one of the inventors of Quickcheck, to discuss adoption strategies, how to gauge the accuracy and effectiveness of property tests, potential ideas for the future, and any questions the people in the audience might have.

Thursday evening keynote at the Code BEAM SF where you can hear all about the newly formed Erlang Ecosystem Foundation. A talk from Jose Valim, Peer Stritzinger, Fred Hebert, Miriam Pena and Francesco Cesarini who are sharing the journey and goals of the foundation that we've all waited for!

The Erlang Ecosystem Foundation's goal is to grow and support a diverse community around the Erlang and Elixir ecosystem, encouraging the continued development of technologies and open source projects based on/around its runtime and languages.

Any system that is successful necessarily grows more complex. This means that code gets messier, but also that the people who are part of the system have to handle ever-increasing complexity.

It is not sufficient to take a code-centric approach; to make our Erlang and Elixir systems truly operator-friendly, we have to understand how our mental models work, and what constitutes good automation. Finally, we need to be aware of all the tools the Erlang VM makes available to us to truly deal with the unexpected.

OBJECTIVES

• Introduce some concepts to think about complexity in large systems and how humans cope with it
• Give measures of what makes a good or a bad abstraction in complex systems; what to automate and when to do so
• Present some of the tools that the Erlang virtual machine provides to deal with big surprises and point to ways to introduce them to your regular workflow

TARGET AUDIENCE

People nervous about having to operate and debug the Erlang and Elixir systems they put in production, or curious about the ways in which they could improve that experience.

Bugs you know you will happen in production are usually easy to handle since you know about them. The problem is that we, as humans, have limited knowledge and understanding about the problem domains we write programs for and very often about our programs themselves. This is at the root of all the surprising unexpected bugs that we actually encounter all the time. While we can become experts at being careful and well-equipped in tooling and just knowing all kinds of weird situations that can arise from code and computers being a thing that we all have to live with, there are some other approaches that can prove helpful as well.

In this talk, we'll see some broad and diverse approaches to handling the unexpected, but will keep a specific focus on how building the proper supervision structures lets us encode predefined failure semantics into our programs so that we know how most errors will be handled, without even knowing what they are.