> Its not a fault of the tests that they fail on wine, its a fault of
> wine. For my understanding the test are created to show that either the
> behaviour of wine is correct (when the test succeeds) or that something
> left to do one wine (when a test fails)!?

check out the todo_wine construct in winetest

Alexandre point is not that you cannot add a test that currently fails in Wine. His point is that you cannot add a test that will make the test process fail. The todo_wine is done for that:

1/ it marks a test as failing on wine

2/ it will report it (but not make the test fail)

3/ it will also make the test process fail if the test succeeds (so that we can remove the todo_wine mark)

A+