Maintaining Automated Browser UI Testing

Question

What's the best way to manage a slew of browser UI tests? I'm looking for an approach that may have worked for you in the past when dealing with numerous automated browser tests. Obvious answers such as "they should be refactored into lower-level UI tests" aren't what I'm looking for. Ultimately these tests are incredibly time consuming to both run and maintain. I'm looking for the best ways to minimize this problem.

I must also mention that I'm confined to free software that's particularly focused on .NET (WatiN, CC.net, Fitnesse, etc.).

Answer 1

Factor out common sequences of actions and just put them in one subtest each. Run only the subtests you need for any given feature.

Answer 2

In other threads about Web UI testing, Selenium was a popular and highly recommended choice.

Answer 3

Selenium is good, but kindof slow. Since unit tests should rarely (if ever) go out of process, they should be much faster than your selenium tests.

If you do use it for automated UI testing, I would put them all in a separate build which can run in parallel with your unit tests.

I prefer to treat my automated UI tests as regression tests.