Here's what we currently do for a click test: https://github.com/thoughtbot/capybara-webkit/blob/master/src/WebPage.cpp#L286

We could additionally check to see that the coordinates we're trying to click are on the screen (or can be scrolled to).

We also check that the node is visible in this code:

@jferris Yes - it currently errors correctly if one tries to click the element - the issue (if there is an issue) is that if #visible? is called on the element it returns true. I'm not sure whether or not that should be the case since the user can't usually scroll to negative positions to see the element.

Note: this only came up because of the :allow_label_click option for checkboxes/radio buttons in Capybara 2.8+ which capybara-webkit doesn't yet support, and I'm working around it for a 2.9.1 release, but it would be nice to have a consensus on what the result of #visible? should be for these elements.

@mhoran if we used clickTest when asking if an element was visible, that would change the original example to return false, right?

Yeah, I think so. Though something may be visible, but underneath another element, in which case clickTest would return false. So I guess we should just add bounds checking to isNodeVisible.

Hmm, actually, that's interesting: if something is obscured behind a modal dialog or something, is it visible?

I believe selenium considers something visible if the user can scroll it into a position where it can be seen, so behind a modal could still be considered visible, of course that may be changing as selenium moves towards only supporting browsers through vendor provided drivers, and the webdriver spec solidifies.

@jferris I was also thinking something along the lines of an overlaying semi-transparent (or completely transparent div), that allows the underling content to be visible, but not clickable.