Achieving peace of mind through test coverage js-montreal 10/2019

Ruy Adorno @ruyadorno OSS Dev @

No content

Intro Tests are an essential tool that helps alleviate the task of maintaining a code base. They give us confidence when shipping new releases of a program by providing an automate assessment into its expected set of features. My hope with this talk is to enlighten on how coverage of code is important and provide practical tips on how to test more of your code.

What is a test? • Program that runs your code • Validates your expectations • Makes sure things behave the way they should

Benefits: • Reliability, you want to know about defects as soon as possible • Approachability, people can tweak your code and know if they break it • Quality, teams end up finding improvements and optimizations

Coverage • Metric, it tracks what parts of your program actually executed • Doesn’t prove that the test suite is good • That can serve as a test itself

Benefits: • In order to achieve higher coverage you have to dig deeper into the internals • Help to uncover bugs, serves as documentation to all program details • Serves as confidence meter for that code

Mergebot • Github bot • Controls merging PRs • Front End cli scripts

• Let’s go through some examples of helpful patterns while trying to bump test coverage Useful patterns:

Snapshots

Snapshots • Useful when testing verbose output • pros: Extra convenient to work with • cons: requires way more discipline

Patch APIs

Patch APIs • Allow us to assert for specific internal calls (specially global)

Abstractions

Abstractions • Abstract hard to test internals into separated function/modules

Mocking modules

Mocking modules • Cleaner solution to checking to internal calls • Test frameworks tend to provide simpler to use APIs for those

node-tap

node-tap • Different test frameworks provide a different set of APIs but conceptually they all share the same ideas

node-tap

node-tap • One very convenient design choice of node-tap is allowing to run tests without the need of a top-level runner • Makes it so much easier to start debugging your tests

It may be difficult to get coverage up, specially when reaching to parts of your codebase in which are harder to test by the very nature of the code there. I hope the patterns I share here help you on the task of covering more features of your code base using automated tests. Conclusion:

process.exit(0) https://ruyadorno.com Thanks!

Photo Credits • Photo by Kari Shea on Unsplash • Photo by Grzegorz Mleczek on Unsplash • Photo by rawpixel on Unsplash • Photo by Aidan Hancock on Unsplash