[PATCH] dmstyle/tests: Relax style descriptor test with nothing loaded

[Alex Henrie]

Signed-off-by: Alex Henrie 
---
This test has been failing on the Windows 7 testbot since February 26,
likely because of a Windows update.

dlls/dmstyle/tests/dmstyle.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/dlls/dmstyle/tests/dmstyle.c b/dlls/dmstyle/tests/dmstyle.c
index ed12315a19..a773607f8b 100644
--- a/dlls/dmstyle/tests/dmstyle.c
+++ b/dlls/dmstyle/tests/dmstyle.c
@@ -458,9 +458,8 @@ static void test_parsedescriptor(void)
     /* Nothing loaded */
     hr = IDirectMusicObject_GetDescriptor(dmo, &desc);
     ok(hr == S_OK, "GetDescriptor failed: %08x, expected S_OK\n", hr);
-    ok(desc.dwValidData == DMUS_OBJ_CLASS /* XP */ ||
-            broken(desc.dwValidData == (DMUS_OBJ_OBJECT | DMUS_OBJ_CLASS)), /* Vista and above */
-            "Got valid data %#x, expected DMUS_OBJ_OBJECT\n", desc.dwValidData);
+    ok(desc.dwValidData & DMUS_OBJ_CLASS,
+            "Got valid data %#x, expected DMUS_OBJ_CLASS\n", desc.dwValidData);
     ok(IsEqualGUID(&desc.guidClass, &CLSID_DirectMusicStyle),
             "Got class guid %s, expected CLSID_DirectMusicStyle\n",
             wine_dbgstr_guid(&desc.guidClass));

[Michael Stefaniuc]

On 03/15/2018 06:11 AM, Alex Henrie wrote:
> Signed-off-by: Alex Henrie
>
> This test has been failing on the Windows 7 testbot since February 26,
> likely because of a Windows update.

Heh, that's when I added that test:
git log 9ae8b8c00f2c dlls/dmstyle/

I noticed the issue two days later and tried to reproduce it with
TestBot job 36301 (already deleted on the testbot). I couldn't!
Here's are reruns with just a whitespace change:
https://testbot.winehq.org/JobDetails.pl?Key=36793
https://testbot.winehq.org/JobDetails.pl?Key=36794
https://testbot.winehq.org/JobDetails.pl?Key=36795

Consistently not reproducible when run manually.
Consistently reproducible when run as part of the daily run but *only*
on newtb-w7pro64-32: The returned data is not random, always an extra
DMUS_OBJ_VERSION that shouldn't be there.

I've compared the dll version and it is on all the exact same dmstyle
6.1.7600.16385

That's why I asked back then Francois:
[23:37:06] <puk> fgouget_: Tag: newtb-w7pro64-32
[23:37:14] <puk> fgouget_: which box is that?
[23:37:37] <puk> w7pro64 (32 bit) ?
[23:38:04] <puk> because I don't see test failures when running the
tests manually on the WTB
[23:38:17] <puk> newtb-w7pro64-32 is the only one to show it
[23:59:35] <fgouget_> puk: it's a test VM, I think you can ignore it as
I will take it down soon
[23:59:52] <fgouget_> set as in for testing the TestBot, not for testing
Wine
[23:59:56] <fgouget_> s/set/test/

So I'm ignoring that issue as it will go away.

bye
    michael
> dlls/dmstyle/tests/dmstyle.c | 5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
>
> diff --git a/dlls/dmstyle/tests/dmstyle.c b/dlls/dmstyle/tests/dmstyle.c
> index ed12315a19..a773607f8b 100644
> --- a/dlls/dmstyle/tests/dmstyle.c
> +++ b/dlls/dmstyle/tests/dmstyle.c
> @@ -458,9 +458,8 @@ static void test_parsedescriptor(void)
>      /* Nothing loaded */
>      hr = IDirectMusicObject_GetDescriptor(dmo, &desc);
>      ok(hr == S_OK, "GetDescriptor failed: %08x, expected S_OK\n", hr);
> - ok(desc.dwValidData == DMUS_OBJ_CLASS /* XP */ ||
> -
>        broken(desc.dwValidData == (DMUS_OBJ_OBJECT | DMUS_OBJ_CLASS)), /* Vista and above */
> -
>        "Got valid data %#x, expected DMUS_OBJ_OBJECT\n", desc.dwValidData);
> - ok(desc.dwValidData & DMUS_OBJ_CLASS,
> -
>        "Got valid data %#x, expected DMUS_OBJ_CLASS\n", desc.dwValidData);
> ok(IsEqualGUID(&desc.guidClass, &CLSID_DirectMusicStyle),
>            "Got class guid %s, expected CLSID_DirectMusicStyle\n",
>            wine_dbgstr_guid(&desc.guidClass));

[Alex Henrie]

Missatge de Michael Stefaniuc  del dia dj., 15 de març
2018 a les 13:40:
> On 03/15/2018 06:11 AM, Alex Henrie wrote:
> > Signed-off-by: Alex Henrie
> >
> > This test has been failing on the Windows 7 testbot since February 26,
> > likely because of a Windows update.
>
> Heh, that's when I added that test:
> git log 9ae8b8c00f2c dlls/dmstyle/
> I noticed the issue two days later and tried to reproduce it with
> TestBot job 36301 (already deleted on the testbot). I couldn't!
> Here's are reruns with just a whitespace change:
> https://testbot.winehq.org/JobDetails.pl?Key=36793
> https://testbot.winehq.org/JobDetails.pl?Key=36794
> https://testbot.winehq.org/JobDetails.pl?Key=36795
> Consistently not reproducible when run manually.
> Consistently reproducible when run as part of the daily run but *only*
> on newtb-w7pro64-32: The returned data is not random, always an extra
> DMUS_OBJ_VERSION that shouldn't be there.
> I've compared the dll version and it is on all the exact same dmstyle
> 6.1.7600.16385
> That's why I asked back then Francois:
> [23:37:06] <puk> fgouget_: Tag: newtb-w7pro64-32
> [23:37:14] <puk> fgouget_: which box is that?
> [23:37:37] <puk> w7pro64 (32 bit) ?
> [23:38:04] <puk> because I don't see test failures when running the
> tests manually on the WTB
> [23:38:17] <puk> newtb-w7pro64-32 is the only one to show it
> [23:59:35] <fgouget_> puk: it's a test VM, I think you can ignore it as
> I will take it down soon
> [23:59:52] <fgouget_> set as in for testing the TestBot, not for testing
> Wine
> [23:59:56] <fgouget_> s/set/test/
> So I'm ignoring that issue as it will go away.

Understood, thanks for the explanation. If setting DMUS_OBJ_VERSION is not
necessarily wrong, it might be a good idea to take the patch anyway. At the
very least, the patch improves the failure message to match what was
actually tested. But I defer to the sole guru of DirectMusic ;-)

-Alex

[Michael Stefaniuc]

On 03/16/2018 05:40 AM, Alex Henrie wrote:
> Missatge de Michael Stefaniuc  del dia dj., 15 de març
> 2018 a les 13:40:
> > On 03/15/2018 06:11 AM, Alex Henrie wrote:
> > > Signed-off-by: Alex Henrie
> > >
> > > This test has been failing on the Windows 7 testbot since February 26,
> > > likely because of a Windows update.
> >
> > Heh, that's when I added that test:
> > git log 9ae8b8c00f2c dlls/dmstyle/
> > So I'm ignoring that issue as it will go away.
>
> Understood, thanks for the explanation. If setting DMUS_OBJ_VERSION is not

It is not setting that, it fresh objects are created with that outside
the control of the app.
> necessarily wrong, it might be a good idea to take the patch anyway. At the

It is more for documenting inconsistencies between the different
IDirectMusicObject implementations. So I'd rather add another broken or so.
> very least, the patch improves the failure message to match what was
> actually tested. But I defer to the sole guru of DirectMusic ;-)

My gut feeling tells me something else is wrong, some unitialized memory
or maybe reuse of a previous object, who knows. The non-reproducible
part during manual test runs makes it very annoying to troubleshoot.
There's some time until Jeremy comes after me at WineConf so I'd rather
keep it at least until I get to cleanup dmstyle.

bye
    michael