(Actually) continuous testing

The term continuous testing typically means building the code and creating a test environment for every version-control change, then running all the types of tests against it – unit, integration, API, and GUI. In pseudo-code, which looks like this:

while (<CT server is on> {
Wait for a change
Check out the code
Perform a build, report errors
          - Create build artifacts
- Run unit tests, report errors
Create a test environment
Run API tests, report errors
Run end to end tests, report errors
}

One of our reviewers, Stephen Spady, pointed out that continuous testing could include post-release, and running synthetic transactions against the live system. We don’t always see this in practice (notice the word typically in the previous paragraph) – but it would be a better world if we did. That would force end-to-end checks to be structured so they could run in production...