Problem: We can't enable autotests on new platforms because a lot of the tests are failing. Currently the solution is to create bug reports out of each and every failure we find, perhaps blacklist the lot and hope for a fix. This is very slow and cumbersome.

Solution: We will add a new feature flag in platform_configurations in qt5.git, say "IgnoreTestFailures". This will pass on the information to the agent not to mind about the exit code of the autotests. If a build now has these failing autotests, we will not mark it as green, but instead use the color brown or something. It's like "yes it passed, but...". When such a build eventually passes without failures, the build could leave a remark that the "ignoretestfailures" should be removed. Where this remark is done or how Coin reports this is still open.