Hi, I've tried addressing the problem of detecting javascript errors during selenium test runs. The solution I found is firefox specific (plugin based), but does expose a couple of issues in the 5.4 tests.
Using the plugin + a trivial testng listener it is possible to add an extra check for javascript console errors to every successful selenium test method execution. And change the test success to a failure, in case errors are detected. As we all know javascript console errors aren't always that descriptive, but this should be enough to be able to reproduce the errors and fix them, as it uses the normal testng reporting to inform about where the errors happened. I hope this (or something similar/even better - feel free to pitch in) can be used for shaking out the last obvious T5.4 client side errors exposed by the current test suite... If the proposed code in https://issues.apache.org/jira/browse/TAP5-2239cannot be incorporated into the current test suite due to licenses (the ff plugin is ASL2 licenced) or something else, it is my hope that it can at least be used to fix some errors exposed by doing a local test run. -- Chris
