“Learning how your app goes ‘boom’ in the real world.”

  by James Turner

Abstract

Your application is feature complete, you fixed all the bugs, and your unit-tests pass! Time to ship it and get on with your life. Unfortunately, then you start to get reports that it crashes for some people, sometimes, on some machines. But they can’t remember what they were doing. Or know what version they were running. Or whether they’re running macOS or Windows.

In this session, we’ll look at my experiences adding real-time automated crash- and error-reporting to a large open-source application, and what I learned in the process. We’ll consider correlating manual user feedback (bug reports) with automated feedback, and adding tracing to our application workflow to understand the steps leading up to an error.

We’ll also review the build and tooling changes required for crash reporting, and integration with other systems, and some other interesting metrics we obtain from the system.

About the speaker

James is a senior software developer at KDAB, and a principal maintainer of FlightGear, the open-source flight simulator. He wrote his first software on the original Macintosh in the mid-90s, and since then has developed and maintained C++ applications on the desktop using a wide range of operating systems and user-interface technologies.