Introduction
At the BBC we build a large number of TV and mobile applications. In order to ensure the widest possible audience reach we have to test these applications across hundreds of different device models.
This is enormously expensive and slows down our abilty to push new features out to our audiences. In order reduce the manual testing effort, we invested heavily in automation and became very good at writing automated tests to help alleviate the testing effort. This only helped so much; although we had good automation suites, actually running them was an expensive manual job – locating devices, setting up test environments, executing the scripts, and collating and interpreting the results.
Hive CI was born out of a desire to entirely eliminate the manual effort of running our automated tests. Conventional CI platforms weren’t really a good fit for what we were trying to do. We wanted a CI system that could understand and manage devices, run our tests without manual intervention, and collate and interpret results without having to resort to spreadsheets.