Undercover: Coverage Lies

Devcon June Conference, Thursday, June 20, 2013, 13:30

For years we’ve looked for coverage as the key metric representing quality. We know it’s not much, but it’s what we’ve got, right?

The truth is that coverage lies all the time. The easiest way to trick coverage tool is remove the ”assert” statement from the test. There are a couple of ways we’re tricking ourselves to believe the coverage report, painting the wrong picture.

In this session, I’ll go over what we believe are coverage truths (usually made to believe by tool vendors…) and what to actually look for in coverage. Coverage reports are not useless – you just need to know what to look for.

We build assumptions and make decisions based on what coverage tells us. Let’s make sure we start by understanding what real useful coverage is.

Gil Zilberfeld has been in software since childhood, writing BASIC programs on his trusty Sinclair ZX81. With more than twenty years of developing commercial software, he has vast experience in software methodology and practices

 

Gil is an agile consultant, applying agile principles over the last decade. From automated testing to exploratory testing, design practices to team collaboration, scrum to kanban, and lean startup methods – he’s done it all. He is still learning from his many failures

Gil speaks frequently in international conferences about unit testing, TDD, agile practices and communication. He blogs at http://www.gilzilberfeld.com and in his spare time he shoots zombies, for fun

Other Presentations at Devcon June