[jolt]

Haven't used it myself, but i would love to hear from someone who has used this, or something similar. How useful is this kind of testing? I'm thinking that tests would have to be re-recorded a lot, if this is used while developing the site.

[hugs]

It sounds like their approach is the same one we took with the first version of Selenium. The approach is JavaScript injection through an HTTP proxy. Because of the same origin policy and JavaScript security model, we abandoned this approach when moving to Selenium 2. Selenium 2 goes for the most native approach for clicks and key events. (Continuing in that native-is-better approach, I'm now experimenting with robots, but that's another story.) I'm amazed they think their approach is better.

I'm surprised they think the Selenium 1 JavaScipt proxy-injection model is a better approach. But hey, I'm getting old... Get off my lawn, yadda, yadda, yadda. But I'm stunned they think it's better and trying to patent it. I really hope I'm missing some new key insight here. Please enlighten me, people.

[jolt]

Thank you. From what i have gathered here in the comments, it seems that I need to look into Selenium.

[janekk]

I'm in the 'something similar' camp. This is functional testing. I use Selenium, an open source library for automating browsers via plugins.

Personally I find it much more useful than unit/integration tests because while those test the programmer's understanding of his own code, functional tests actually drive the app, perform normal user actions by driving the GUI. For example, you could have all of your unit tests passing but without functional tests, your OK button may appear off-screen and you wouldn't know.

It is annoying when you make changes to your app because test runs take much longer than unit tests, so I have one general test that I run after a refactor, and then a full suite that runs after commit.