Finding out that a particular app breaks because of wrong error code might be taking hours quite often, days in the bad case. I was under impression accurately testing the error codes is a common Wine practice, unless it is too complicated. I think adding NULL parameters test is something we used to have in most of functions tests (not all the possible combinations of nulls of course). So if while adding the test we found that some null causes crash, why should we spare a line or two to reflect this with a comment and leave the reader to guess whether it was not tested ir crashing on some Windows versions?
On 2 Sep 2021, at 12:13, Huw Davies huw@codeweavers.com wrote:
Of course if a particular app depends on Windows returning a specific error code then thats the time to add a test for it.
Huw.