I have one problem with ReactOS

[w3seek]

You do NOT accept reverse engineered original code in the ROS source, right?

That's right, it's not appreciated

Do you accept and use specification of original code behaviour that was obtained from disassembling original code? Or do you only use data from "test runs" of original code?

Hard to say, generally data from test runs is preferred. As long as the code was implemented without reverse-engineering but the knowledge of it's behavior it should be ok.

Do you accept and use input data for important test runs, i.e. that test run and its output data give you important clues of the original code behaviour, even if the importance of that data/test run was seen by disassembling original code?

As mentioned above, this is much more appreciated. These test runs are also useful for regression testing.