[Bug 28491] New: Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

Summary: Gangsters 2 Demo (German): Video test won't run
           Product: Wine
           Version: 1.2
          Platform: x86
               URL: http://www.pcgames.de/Gangsters-2-PC-16428/Downloads/G
                    angsters-2-spielbare-Demoversion-9842/
        OS/Version: Linux
            Status: UNCONFIRMED
          Severity: enhancement
          Priority: P2
         Component: user32
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]
                CC: [e-mail]
   Regression SHA1: f1ea06375d76c9f8e181e3ee4874ddef661e80c2

Created attachment 36545
  --> http://bugs.winehq.org/attachment.cgi?id=36545
An image you see when the video test is running

There is a regression in device context implemetation.

When you run the game for the first time you should see an image like the one
attached here.

Regression test told

f1ea06375d76c9f8e181e3ee4874ddef661e80c2 is the first bad commit
commit f1ea06375d76c9f8e181e3ee4874ddef661e80c2
Author: Alexandre Julliard 
Date:   Thu Jul 22 20:00:35 2010 +0200

user32: DCX_NORESETATTRS should be applied at DC release time.

:040000 040000 dec10ae5d778f859fcc46984e0a01e2abbfba25b
8ffe15178f0ef46a091fed411153ab9e6da8cc36 M    dlls

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

Bruni  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |download, regression
            Version|1.2                         |1.3.0

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #1 from Bruni  2011-09-25 06:53:42 CDT ---
Created attachment 36546
  --> http://bugs.winehq.org/attachment.cgi?id=36546
trace with wine-1.3.0

WINEDEBUG=+tid,+seh,+relay ~/wine-git/wine explorer /desktop=name,800x600
./Gangsters2Demo.exe &> log.txt

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #2 from Alexandre Julliard  2011-09-25 14:42:22 CDT ---
This looks like leftover garbage, more like a bad window refresh. The patch
doesn't seem to have anything to do with this except change the timings. Do you
really need to see this stuff?  Do you see it on Windows?

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #3 from Bruni  2011-09-27 00:54:59 CDT ---
That was actual data gotten during video test. Windows XP shows it and so did
Wine prior to 1.3.0. Perhaps the patch simply don't give time enough to reveal
that data.

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #4 from Bruni  2011-12-25 03:40:10 CST ---
somewhere between 1.3.28 and 1.3.33 it was fixed but as of
wine-1.3.35-168-g5b93bb9 the bug is back

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #5 from Alexandre Julliard  2011-12-25 04:49:21 CST ---
Please do a new regression test then.

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

Bruni  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]
    Regression SHA1|f1ea06375d76c9f8e181e3ee487 |5de020342cc039819b6fe11ee78
                   |4ddef661e80c2               |7d5a9a288f85a

--- Comment #6 from Bruni  2012-01-05 04:50:15 CST ---
5de020342cc039819b6fe11ee787d5a9a288f85a is the first bad commit
commit 5de020342cc039819b6fe11ee787d5a9a288f85a
Author: Henri Verbeet 
Date:   Tue Sep 13 20:02:25 2011 +0200

ddraw: Get rid of the unused "initialized" field in IDirectDrawImpl.

:040000 040000 4b4589f852778974d9543e4e37f728ad04099513
83a12a7138c35cec2bf005afb86af8e1dba21a14 M    dlls

CC-ing Henri

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #7 from Henri Verbeet  2012-01-05 05:29:01 CST ---
That patch did cause a regression at the time, but
ec04f80cf606f07a630bfc515f004be9cc4679ff should have fixed it again.

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #8 from Bruni  2012-01-05 06:32:01 CST ---
ec04f80cf606f07a630bfc515f004be9cc4679ff dates back to September, 18, whereas
the regression is present as of 1.3.36.
Is there any chance that ec04f80cf606f07a630bfc515f004be9cc4679ff envelops not
all regressions caused by 5de020342cc039819b6fe11ee787d5a9a288f85a ?

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #9 from Henri Verbeet  2012-01-05 06:53:27 CST ---
(In reply to comment #8)
> ec04f80cf606f07a630bfc515f004be9cc4679ff dates back to September, 18, whereas
> the regression is present as of 1.3.36.
> Is there any chance that ec04f80cf606f07a630bfc515f004be9cc4679ff envelops not
> all regressions caused by 5de020342cc039819b6fe11ee787d5a9a288f85a ?

Maybe, but I'd be somewhat surprised. Does it work with
ec04f80cf606f07a630bfc515f004be9cc4679ff, or even just
ec04f80cf606f07a630bfc515f004be9cc4679ff applied on top of
5de020342cc039819b6fe11ee787d5a9a288f85a?

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #10 from Bruni  2012-01-05 07:21:13 CST ---
The image of video test doesn't show up with
ec04f80cf606f07a630bfc515f004be9cc4679ff, but it is fine just before
5de020342cc039819b6fe11ee787d5a9a288f85a
I didn't understand just what you meant under
"ec04f80cf606f07a630bfc515f004be9cc4679ff applied on top of
5de020342cc039819b6fe11ee787d5a9a288f85a"

I have one thought. What if a buggy commit of some third author was committed
between 5de020342cc039819b6fe11ee787d5a9a288f85a and
ec04f80cf606f07a630bfc515f004be9cc4679ff?

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #11 from Henri Verbeet  2012-01-05 07:38:51 CST ---
(In reply to comment #10)
> I didn't understand just what you meant under
> "ec04f80cf606f07a630bfc515f004be9cc4679ff applied on top of
> 5de020342cc039819b6fe11ee787d5a9a288f85a"

Checkout 5de0203, then apply ec04f80 on top of that. E.g. by cherry picking it
or doing something like "git show ec04f80 | git apply -".
> I have one thought. What if a buggy commit of some third author was committed
> between 5de020342cc039819b6fe11ee787d5a9a288f85a and
> ec04f80cf606f07a630bfc515f004be9cc4679ff?

Yeah, that's what I wanted to test with applying ec04f80 on top of 5de0203.

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #12 from Bruni  2012-01-06 10:32:41 CST ---
> Checkout 5de0203, then apply ec04f80 on top of that. E.g. by cherry picking it
> or doing something like "git show ec04f80 | git apply -".

What I did:
git checkout 5de020342cc039819b6fe11ee787d5a9a288f85a
git cherry-pick ec04f80cf606f07a630bfc515f004be9cc4679ff
./configure && make

-> video test works fine

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #13 from Henri Verbeet  2012-01-06 10:44:18 CST ---
It's probably one of the commits in between then. You can still bisect that,
but it's a bit tedious because you have to apply ec04f80 after each step.

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #14 from Bruni  2012-01-06 11:11:05 CST ---
> It's probably one of the commits in between then. You can still bisect that,
> but it's a bit tedious because you have to apply ec04f80 after each step.

Could you please explain this to me, why ec04f80 needs cherry-picking each time
I move through git tree? Why can it not be so that ec04f80 takes effect for all
subsequent commits in git once and for all?

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #15 from Henri Verbeet  2012-01-06 11:32:30 CST ---
Well in principle you could probably do something like the following:

git branch bisect ec04f80
git checkout -b base 5de0203
git cherry-pick ec04f80
git checkout bisect
git rebase base
git branch -D base

and then bisect between HEAD and 5de0203. It's a bit more complicated though,
and your commit ids won't match anymore, so after you're done with bisecting
you'd then have to find the corresponding commit again in the original history.

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #16 from Bruni  2012-01-06 14:38:21 CST ---
> git branch bisect ec04f80
> git checkout -b base 5de0203
> git cherry-pick ec04f80
> git checkout bisect
> git rebase base
> git branch -D base

I'm a novice in git, but I think that doing "git rebase base" and than "git
branch -D base" is wrong. Both ec04f80 and 5de0203 are not present in "bisect"
by the time you delete "base". Instead, is this any good to do

git branch bisect ec04f80
git checkout -b base 5de0203
git cherry-pick ec04f80
git checkout bisect
git merge base
git bisect start
git bisect good 5de0203
git bisect bad HEAD
... etc
git branch -D bisect

What do you think?

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #17 from Bruni  2012-01-07 06:45:09 CST ---
I found it. The problematic commit is
http://source.winehq.org/git/wine.git/commit/1aeee59bd8b6172c848c8029e1a55dd...

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

Bruni  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
    Regression SHA1|5de020342cc039819b6fe11ee78 |1aeee59bd8b6172c848c8029e1a
                   |7d5a9a288f85a               |55ddac1f93c44

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #18 from Henri Verbeet  2012-01-13 12:01:24 CST ---
afb64506ba0eb958af72d8535d8c5473f7fcc06e should fix the video test.

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #19 from Bruni  2012-01-14 11:17:21 CST ---
fixed for as of 1.3.37
Also, the game is now partly playable.

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

--- Comment #20 from Bruni  2012-01-14 11:19:19 CST ---
fixed for me as of 1.3.37
Also, the game is now partly playable.

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

Henri Verbeet  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
      Fixed by SHA1|                            |afb64506ba0eb958af72d8535d8
                   |                            |c5473f7fcc06e
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |FIXED

--- Comment #21 from Henri Verbeet  2012-01-14 14:17:25 CST ---
Resolving.

[Bug 28491] Gangsters 2 Demo (German): Video test won't run

http://bugs.winehq.org/show_bug.cgi?id=28491

Alexandre Julliard  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #22 from Alexandre Julliard  2012-01-27 14:17:43 CST ---
Closing bugs fixed in 1.4-rc1.