On Wed, Mar 21, 2018 at 2:47 AM Huw Davies <huw(a)codeweavers.com> wrote:
On Sun, Mar 18, 2018 at 11:26:21PM -0600, Alex Henrie wrote:
Signed-off-by: Alex Henrie <alexhenrie24(a)gmail.com> --- The failing tests lack permissions or use invalid combinations of flags, so it's not too surprising that the behavior changed.
dlls/user32/tests/broadcast.c | 50 ++++++++++++++++++++++++++++--------------- 1 file changed, 33 insertions(+), 17 deletions(-)
diff --git a/dlls/user32/tests/broadcast.c b/dlls/user32/tests/broadcast.c index afdb536816..851c47eb96 100644 --- a/dlls/user32/tests/broadcast.c +++ b/dlls/user32/tests/broadcast.c @@ -131,8 +131,9 @@ if (0) /* TODO: Check the hang flags */ recips = BSM_APPLICATIONS; ResetEvent(hevent); ret = broadcast( BSF_POSTMESSAGE|BSF_QUERY, &recips, WM_NULL, 100, 0 ); - ok(ret==1, "Returned: %d\n", ret); - ok(WaitForSingleObject(hevent, 0) != WAIT_TIMEOUT, "Asynchronous message sent instead\n"); + ok((ret == 1 && WaitForSingleObject(hevent, 0) != WAIT_TIMEOUT) || + (ret == 0 && WaitForSingleObject(hevent, 1000) == WAIT_TIMEOUT) /* win10 */, + "Asynchronous message sent instead of synchronous message or no message. Returned: %d\n", ret); PulseEvent(hevent);
As you say, some of these flag combinations are meaningless. I suggest we just remove them.
Hmm, I generally assume that test cases are there for a reason, but I'm happy to just remove the six invalid flags tests if there is general consensus that that's the right thing to do. Does anyone else have an opinion? -Alex