October 2009 - Wine-Devel - list.winehq.org

Re: comctl32: implement LISTVIEW_ApproximateViewRect for icon and small icon.
by Nikolay Sivov Oct. 2, 2009

Oct. 2, 2009

Aric Stewart wrote: > --- > dlls/comctl32/listview.c | 39 ++++++++++++++++++++++++++++++++++----- > 1 files changed, 34 insertions(+), 5 deletions(-) > Hi, Aric. I don't think it's a right way: --- + if (infoPtr->nItemCount > 0) + { + LISTVIEW_GetItemBox(infoPtr, 0, &rcBox); + if (wWidth == 0xffff) + wWidth = infoPtr->rcList.right - infoPtr->rcList.left; + + row_count = wWidth / (rcBox.right - rcBox.left); + if (row_count == 0) + { + row_count = 1; + wWidth = rcBox.right - rcBox.left; + } + rows = infoPtr->nItemCount / row_count; + if (rows == 0) + rows = 1; + + wHeight = (rcBox.bottom - rcBox.top) * rows; + } --- You don't respect item spacing here while native does: just tested with ControlSpy and it does return rectangle with item spacing - item aren't placed close to each other. I think you should use a logic we use to position items - place to row until it wraps to next etc. The difference is that if width is greater by a value less then itemwidth we should return a minimal rectangle here. Using LISTVIEW_GetItemBox(infoPtr, 0, &rcBox) for 0 item assuming it's the same for the rest isn't correct - box depends on label so, it could differ. Also this is unsafe: --- + row_count = wWidth / (rcBox.right - rcBox.left); --- Could add some tests for that? Not so strict maybe, but to check that approx. rectangle is wider then item placed close.

1 0

Re: [1/2] user32/tests/cursoricon.c: DrawState: New Testcase for correct drawing of Icons
by Alexandre Julliard Oct. 2, 2009

Oct. 2, 2009

Wilfried Pasquazzo <wilfried.pasquazzo(a)gmail.com> writes: > [1/2] New Testcase to check if DrawState() behaves correctly when drawing Icons. It doesn't work here: ../../../tools/runtest -q -P wine -M user32.dll -T ../../.. -p user32_test.exe.so cursoricon.c && touch cursoricon.ok cursoricon.c:1290: Test failed: Drawing Icon stretched to 2x2: Pixel at 2,2 should still have background Color FFFFFF, Got C0B0A0 from line 1346 cursoricon.c:1262: Test failed: Drawing Icon with size 1x1: Overlaying Mask 1 on Color A0B0C0 with flags 00000000. Expected a close match to 3F4F5F (modern) or 3F4F5F (legacy). Got FFFFFF from line 1347 cursoricon.c:1298: Test failed: Drawing Icon stretched to 2x2: Overlaying Mask 1 on Color A0B0C0 with flags 00000000. Expected a close match to 3F4F5F (modern) or 3F4F5F (legacy). Got FFFFFF from line 1347 make: *** [cursoricon.ok] Error 3 -- Alexandre Julliard julliard(a)winehq.org

2 1

Re: [5/9] wpp: Handle remaining memory allocation failures [try 2]
by Alexandre Julliard Oct. 2, 2009

Oct. 2, 2009

Matteo Bruni <matteo.mystral(a)gmail.com> writes: > @@ -82,8 +82,7 @@ void *pp_xmalloc(size_t size) > res = malloc(size); > if(res == NULL) > { > - fprintf(stderr, "Virtual memory exhausted.\n"); > - exit(2); > + pp_internal_error(__FILE__, __LINE__, "Virtual memory exhausted.\n"); > } > return res; > } > @@ -96,8 +95,7 @@ void *pp_xrealloc(void *p, size_t size) > res = realloc(p, size); > if(res == NULL) > { > - fprintf(stderr, "Virtual memory exhausted.\n"); > - exit(2); > + pp_internal_error(__FILE__, __LINE__, "Virtual memory exhausted.\n"); Memory allocation failures are not internal errors. -- Alexandre Julliard julliard(a)winehq.org

2 1

Re: [7/9] wpp: Remove the assumption pp_internal_error calls exit() [try 2]
by Alexandre Julliard Oct. 2, 2009

Oct. 2, 2009

Matteo Bruni <matteo.mystral(a)gmail.com> writes: > From 79aebdb612334b9816375abce58c65b90130732a Mon Sep 17 00:00:00 2001 > From: Matteo Bruni <matteo.mystral(a)gmail.com> > Date: Tue, 29 Sep 2009 12:10:24 +0200 > Subject: wpp: Remove the assumption pp_internal_error calls exit() Actually it's OK for it to exit(), though assert() would be a better choice. It's not something that should happen or that we can recover from. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [2/5] d3d9: Test viewports that are bigger than the surface
by Henri Verbeet Oct. 2, 2009

Oct. 2, 2009

> + /* Test a viewport with Width and Height bigger than the surface dimensions > + * > + * TODO: Test Width < surface.width, but X + Width > surface.width > + * TODO: Test Width < surface.width, what happens with the height? > + */ > + memset(&vp, 0, sizeof(vp)); > + vp.X = 0; > + vp.Y = 0; > + vp.Width = 10000; > + vp.Height = 10000; > + vp.MinZ = 0.0; > + vp.MaxZ = 0.0; > + hr = IDirect3DDevice9_SetViewport(device, &vp); > + ok(hr == D3D_OK, "IDirect3DDevice9_SetViewport failed with %08x\n", hr); What does GetViewport() return after a SetViewport() call like that? Please add at least a test for d3d8 as well, and if possible for ddraw. The d3d8 tests are already neglected compared to d3d9, no need to make it worse. > + IWineD3DSurfaceImpl *target = (IWineD3DSurfaceImpl *) stateblock->wineD3DDevice->render_targets[0]; You should probably use context->current_rt there.

2 1

Re: [4/5] WineD3D: SetRenderTarget bypasses stateblock recording
by Henri Verbeet Oct. 2, 2009

Oct. 2, 2009

2009/10/1 Stefan Dösinger <stefan(a)codeweavers.com>: > + /* Finally, reset the viewport as the MSDN states. Tests show that stateblock recording is ignored. > + * the change goes directly into the primary stateblock. > + */ > + This->stateBlock->viewport.Height = ((IWineD3DSurfaceImpl *)This->render_targets[0])->currentDesc.Height; > + This->stateBlock->viewport.Width = ((IWineD3DSurfaceImpl *)This->render_targets[0])->currentDesc.Width; I know you wrote tests for this, could you please submit those as well? "pRenderTarget" should work just as well as "This->render_targets[0]".

2 2

Re: [5/5] WineD3D: SetRenderTarget resets the scissor rect
by Henri Verbeet Oct. 2, 2009

Oct. 2, 2009

Please submit the tests for this one as well.

1 0

Re: [1/2] user32/tests/cursoricon.c: DrawState: New Testcase for correct drawing of Icons
by Nicolas Le Cam Oct. 1, 2009

Oct. 1, 2009

2009/10/1 Wilfried Pasquazzo <wilfried.pasquazzo(a)gmail.com>: > [1/2] New Testcase to check if DrawState() behaves correctly when drawing Icons. > > affected files: dlls/user32/tests/cursoricon.c > > Currently checks: > > - if the Icons are drawn at the correct size. Should always be the > original icon size, > independent of the width and height parameters passed to DrawState(). > - if Icons without a bit-mask are drawn with correct color. > - if Icons with a bit-mask are drawn with correct color. > > Test is successful in WindowsXP SP2, but fails in git-Wine, because > the Icons are > drawn too big. > > ---- > > [2/2] Make Wine pass the new Testcase and Fix Regression Bug 20153 > > affected files: dlls/user32/uitools.c > > Make DrawState() use the Icon dimensions when drawing. This makes it pass all > in [1/2] implemented tests and fixes regression > http://bugs.winehq.org/show_bug.cgi?id=20153 . The width and height, that are > passed to UITOOLS_DrawStateJam() are taken directly from the icon that is to be > drawn (see implementation of UITOOLS_DrawState()), therefore it is correct to > simply use them as the target drawing dimensions. > > Note: This fix was sent before without the testcase (see > http://www.winehq.org/pipermail/wine-patches/2009-September/078927.html) > but not committed. Didn't get feedback on that, so I guessed the reason was the > missing testcase. > > ---- > > Feedback is welcome! > > Wilfried Pasquazzo > > PS: When this gets committed, I can add other tests for DrawState(), > to e.g. test for > correct drawing of the "Disabled" effect. > > > > Hi Wilfried, If test fails on Wine, mark them todo_wine. -- Nicolas Le Cam

1 0

Re: [4/4] msi: Test installing over in-use files.
by Paul Vriens Oct. 1, 2009

Oct. 1, 2009

On 09/29/2009 02:57 PM, Hans Leidekker wrote: > > + static char key[] = "System\\CurrentControlSet\\Control\\Session Manager"; > + static char value[] = "PendingFileRenameOperations"; > + > + RegOpenKeyExA(HKEY_LOCAL_MACHINE, key, 0, KEY_ALL_ACCESS,&hkey); > + if (!RegQueryValueExA(hkey, value, NULL, NULL, NULL,&size)) > + { > + skip("Pending file rename operations, skipping test\n"); > + return; Hi Hans, (In response to this and the previous patch concerning pending renames). Win9x and WinMe act differently on pending renames (as usual :( ). Instead of using the registry some entries are written into %windir%\WinInit.ini. Is it worthwhile to have the tests cope with this or should we just skip them on Win9x/WinMe? -- Cheers, Paul.

2 1

Wine and Intel 3D
by MD.IMAM HOSSAIN Oct. 1, 2009

Oct. 1, 2009

Dear Developers, Latest wine is having problems with Intel UXA 2D driver. Many games cause X server to restart. My system is UBUNTU 9.04. Tested with Intel 2.8 graphics driver, 2.6.30 rc-4 linux kernel, Mesa 7.5.2. WINE 1.1.29 = prolem WINE 1.1.18 = NO problem All Linux native games works wonderfully by this latest Intel driver. Best regards, MD. IMAM HOSSAIN

3 2