Post by Sebastian Hädrich

It should go without saying, but apparently, there are still companies who make the mistake to measure and communicate time to first deployment to production as the primary metric to tell how long a feature takes. But this is misleading: A feature isn’t actually finished until the last bug (sure, there will be more in the future, but “last” as in “it can be used without bugs for now as promised to the customer”) is fixed. So, trying to deploy as fast as possible might make marketing happy, but in the long run, it will hurt you (and probably your reputation as a company).

That being said, there sure are some aspects of #software_development which are in fact optional and #software_testing can cost more time than it saves—when it is done wrong. However, generally speaking, testing, be it #TDD, #BDD, #ATDD, or other approaches, should not be optional. Same goes for #code_quality. Sure, this needs to be done at a reasonable level, perfection doesn’t need to be the goal, but “no matter how bad the code is as long as it is written fast” shouldn’t be it either.