Fantastic Bugs and How to Test Them

Software engineers universally admit that their code is broken. Bugs don’t seem like single differentiated events but rather individual snowflakes in a constant avalanche. Perhaps we should crystallize this into a principle: Code is broken unless it’s proven working.

This principle of course raises the question of how we should prove code works. Tests are the seemingly-obvious answer but given C++’s low level nature we often see proper testing neglected because it would be difficult or it is said to be impossible. However principles are stronger than suggestions: If code is broken until it’s proven working we need to write tests regardless of how difficult it is.

This talk will take the audience on a tour of several real world C++ bugs, the test code written to prove they were addressed, and the structure of the surrounding code which enabled and supported said tests. Unit testing will be shown not to be a luxury to be afforded, but a necessity which cannot be dispensed with.