Only one perspective:
I am working on a web application that is the interface of a video analytics system. (The back panel is usually an IP camera, DVR or video router with a very poor built-in web server.) It uses several jQueryUI widgets, allows the user to configure the device, create video analysis rules and draw markup on top of the video using elements canvas. I find it pretty complicated.
We use unit tests (originally written for JSUnit but now using qunit) for a very limited subset of code. We have unit tests to test the behavior of business objects, including the ability to serialize / deserialize to / from XML. And we have unit tests for testing the basic geometry classes that we wrote for marking the canvas.
However, we do not have unit tests that manipulate the DOM or verify that the elements on the page are in the correct state. Doing this correctly showed us an overly complex problem to solve, so we rely on Selinium tests to verify that this set of inputs will put the DOM in the correct state.
dgvid source share