(I've just spotted this; sorry I'm late to comment) I've seen this issue, on and off, for some time now. In my experience, it's not very predictable, but the hypothesis that it's caused by pages being incompletely loaded is plausible ... although I've experienced this sort of thing when doing a mass of job renames in multiple tabs when I've left a good deal of time between when I originally asked for the page to load and when I tried to use the page. IMO the basic requirement here is that the anti-CRFS functionality should not break things just because a user didn't wait for cosmetic fluff to finish loading - all essential functionality should be in place before the user is allowed to click on anything. I would point out, however, that you don't need to be "fast" to be able to click on things before the page had finished loading; you'd just need to be "faster" than the page loading time, which might actually be very slow, thus allowing a user who's acting at a "perfectly reasonable speed" to be way ahead of their browser. In my experience, with a heavily loaded Jenkins server that's running on a VM that's on a heavily loaded Hypervisor that's accessed over a heavily loaded corporate WAN, I don't need to act "fast" to be faster than the Jenkins UI - the Jenkins UI can be very slow. I guess it's possible that Jenkins was so slow that it might not have finished loading pages despite me waiting... |