July 2009 - Wine-devel - list.winehq.org

Patch feedback (GetProductInfo)
by Alexander Nicolaysen Sørnes 03 Jul '09

03 Jul '09

Hello, Could someone give me a hint as to what is wrong with the following patches? http://www.winehq.org/pipermail/wine-patches/2009-June/074908.html http://www.winehq.org/pipermail/wine-patches/2009-June/074909.html They only adds some defines plus a stub, so hopefully there isn't that much I can have done wrong :) Alexander N. Sørnes

2 2

Re: comdlg32: Fix a problem with the returned value of a CDN_FILEOK notification.
by Paul Vriens 03 Jul '09

03 Jul '09

Rein Klazes wrote: > fixes bug #19079 > > The application of this bug report subclasses the window proc of the file > dialog. It does not return the result in the DWLP_MSGRESULT dialog > property but the return code of the window proc. Tests are provided to > show correct behaviour in all cases, subclassed or not. > --- > dlls/comdlg32/filedlg.c | 8 ++-- > dlls/comdlg32/tests/filedlg.c | 112 +++++++++++++++++++++++++++++++++++++++++ > 2 files changed, 116 insertions(+), 4 deletions(-) > Hi Rein, This one introduced extra test failures on Vista. On 1 Vista box and W2K8 the test times out (crashes?). Could you have a look. -- Cheers, Paul.

2 2

Re: [2/2] gdiplus: Implement GdipBeginContainer2 and GdipEndContainer
by Vincent Povirk 03 Jul '09

03 Jul '09

>+static GpStatus init_container(GraphicsContainerItem** container, >+ GDIPCONST GpGraphics* graphics){ >+ GpStatus sts; >+ >+ if(!container || !graphics) >+ return InvalidParameter; Since this is an internal function and you know container and graphics will always be non-NULL when you call it, you don't need to make this check. >+ sts = GdipDeleteMatrix(graphics->worldtrans); >+ if(sts != Ok) >+ return sts; >+ >+ sts = GdipCloneMatrix(container->worldtrans, &graphics->worldtrans); >+ if(sts != Ok) >+ return sts; If the delete succeeds but the clone fails, you'll have a Graphics with a deleted world transform. You should allocate the objects you need before freeing those attached to the Graphics. I think that in this case you could get away with copying pointers rather than creating new objects, and you can assume the delete functions succeed. Since the container is about to be destroyed, you would need to make sure the attached matrix and region are not destroyed with it. >+ /* did not find a matching container */ >+ if(&container->entry == &graphics->containers) >+ return Ok; That's a little strange, even though the test shows it's not an error. Personally, I would throw in a WARN for this case. I assume you meant to mark this as [1/2] and the test as [2/2]. This looks good to me. It's almost at the point where I can't predict whether it will be accepted (as opposed to being sure it won't), but Alexandre nailed me once for doing a similar "delete before allocating the replacement" thing. So I assume he cares about that. -- Vincent Povirk

1 0

Re: [1/2] gdiplus/tests: Test cases for GdipBeginContainer2
by Vincent Povirk 03 Jul '09

03 Jul '09

+ GdipGetClipBounds(graphics, &clip); + win_skip(/*fabs(defClip[0] - clip.X) < 0.0001 && + fabs(defClip[1] - clip.Y) < 0.0001 && + fabs(defClip[2] - clip.Width) < 0.0001 && + fabs(defClip[3] - clip.Height) < 0.0001,*/ + "Expected Clipping Rectangle to be restored to [%f, %f, %f, %f], got [%f, %f, %f, %f]\n", + defClip[0], defClip[1], defClip[2], defClip[3], + clip.X, clip.Y, clip.Width, clip.Height); Unconditional win_skip (or skip) does not make sense. skip indicates that some tests have been skipped, but this is not the case here. If that fails on Windows, you should fix it. -- Vincent Povirk

1 0

Appdb flight simulation sub category
by Keith Muir 02 Jul '09

02 Jul '09

Hi, Any chance of a games> simulation> flight simulation sub category? Regards, Keith

4 4

Re: fix overflow in several advapi32 crypt functions hidden by broken WideCharToMultiByte
by Dmitry Timoshkov 02 Jul '09

02 Jul '09

"Christoph von Wittich" <Christoph(a)ApiViewer.de> wrote: > + if (*pcbProvName > INT_MAX) > + *pcbProvName = INT_MAX; In which way WideCharToMultiByte is broken? It always helps to provide an explanation and if possible a test case. -- Dmitry.

2 1

Re: user32: Add a test for BM_SETSTYLE, make it pass under Wine.
by Paul Vriens 02 Jul '09

02 Jul '09

Dmitry Timoshkov wrote: > --- > dlls/user32/button.c | 14 ++++++- > dlls/user32/tests/msg.c | 86 ++++++++++++++++++++++++++++++++++++++-------- > 2 files changed, 83 insertions(+), 17 deletions(-) > Hi Dmitry, These new tests introduce some failure on Win98/WinMe. Could you have a look? -- Cheers, Paul.

1 0

Wine64: Down to 68 test failures!
by Austin English 02 Jul '09

02 Jul '09

For those of you that don't follow http://test.winehq.org/, AJ and others 64-bit work has brought down the failures to 68: http://test.winehq.org/data/fb0275dd3148a098967f5959adb57ebe41a4383e/index_… Down quite a bit from over 100 about 6 months ago. There's still a lot of work to do, of course. If you've got a 64-bit machine, give your favorite area of the wine tree a try in 64-bit mode and fix some test failures/compiler warnings. -- -Austin

1 0

Re: regedit: Implement open last key at startup [try 3]
by Michael Karcher 02 Jul '09

02 Jul '09

Am Donnerstag, den 02.07.2009, 17:42 +0500 schrieb Vladimir Pankratov: > Implement open last key at startup. Why do you use wszKey in the get code path, but wszVal in the set code path? But before you change that, read on: +static void get_set_last_key(HWND hwndTV, BOOL bget) This looks to me like the "boolean parameter to change behaviour" antipattern. I think you did it to avoid duplication of the wsz... constants for the registry path and the key name. Make them global variables (still static, so file local), and make two functions out of it. Open the key only for READ access if you want to load. Why do you copy the result of GetItemFullPath? Why don't you free the result of GetItemFullPath? Finally, I think Paul Vriens was wrong on omitting the RegCloseKey. While you really don't have to close predefined keys like HKEY_CURRENT_USER, you do have to close HKCU\Software\Microsoft\Windows \CurrentVersion\Applets\Regedit. Regards, Michael Karcher

2 2

Re: regedit: Implement open last key at startup [try 2]
by Paul Vriens 02 Jul '09

02 Jul '09

Vladimir Pankratov wrote: > Hello. > > Open last key at startup. > > Changed files: > regedit/childwin.c > > Thanks. > > > ------------------------------------------------------------------------ > > Hi Vladimir, Please use a constant instead of hardcoded 260 as suggested by Andre'. You also should be using the RegCreate variant of the function as this key does not yet exist. There is no need for the RegCloseKey as you are opening a predefined one. -- Cheers, Paul.

2 1