June 2011 - Wine-Devel - list.winehq.org

How to do single-step debugging while testing?

by Michael Mc Donnell

Hi I'm trying to figure out how to single step through the code. I've looked at the winedbg and gdb information on winehq, but still can't figure out what to do. I normally run tests by issuing: $ ../../../tools/runtest name_of_test Is there a way to run the test in a debugger, so that I can single step through the functions that the test calls? Thanks, Michael Mc Donnell

June 24, 2011

Re: GSoC-2011: Implement Missing Mesh Functions in Wine’s D3DX9

by Michael Mc Donnell

On Fri, Jun 24, 2011 at 6:12 AM, Dylan Smith <dylan.ah.smith(a)gmail.com> wrote: > On Thu, Jun 23, 2011 at 8:18 AM, Michael Mc Donnell > <michael(a)mcdonnell.dk> wrote: >> I've change the test a bit. I split up the complex example (mesh6) >> into two simpler examples. I also got inspired by your ASCII >> illustrations and have added my own in the comments. The examples have >> also been updated so that they are easier to draw by hand (except for >> mesh5). >> > > Looks good. Great, thanks for the review. I'll sit on them for a little bit more so others can get a chance to comment on them.

June 24, 2011

Re: [PATCH 04/13] winmm: Clean up waveform.c's style

by Paul Vriens

On 06/23/2011 08:12 PM, Andrew Eikum wrote: > + if(dwFlags& WAVE_FORMAT_QUERY) > + TRACE("WAVE_FORMAT_QUERY requested!\n"); How is removing the spaces for all these if/else statements an improvement? It's also not consistent with the rest of the winmm code. -- Cheers, Paul.

June 24, 2011

Re: GSoC-2011: Implement Missing Mesh Functions in Wine’s D3DX9

by Dylan Smith

On Thu, Jun 23, 2011 at 8:18 AM, Michael Mc Donnell <michael(a)mcdonnell.dk> wrote: > I've change the test a bit. I split up the complex example (mesh6) > into two simpler examples. I also got inspired by your ASCII > illustrations and have added my own in the comments. The examples have > also been updated so that they are easier to draw by hand (except for > mesh5). > Looks good.

June 23, 2011

Re: Disablin xinput2 for now?

by Austin English

On Mon, Jun 20, 2011 at 23:26, Vitaliy Margolen <wine-devel(a)kievinfo.com> wrote: > On 06/20/2011 02:31 PM, Austin English wrote: >> >> On Mon, Jun 20, 2011 at 09:36, Alexandre Julliard<julliard(a)winehq.org> >> wrote: >>> >>> Vitaliy Margolen<wine-devel(a)kievinfo.com> writes: >>> >>>> On 06/20/2011 02:55 AM, Marcus Meissner wrote: >>>>> >>>>> As this might be a X.org issue, should we packagers turn off >>>>> xinput2 support for the time being? >>>> >>>> If that will fix all dinput regressions - absolutely. Or add a >>>> registry option to disable xinput2. >>> >>> It will of course fix some regressions, and introduce new ones for >>> things that have been fixed by the changes. Don't tell me that dinput >>> was working perfectly before. And there is a registry option >>> (GrabPointer). >> >> I just did a quick test with today's git (wine-1.3.22-255-g4c0c0d3) >> (which has >> http://source.winehq.org/git/wine.git/commitdiff/adb86c5f2a2584dc8131b08d26…) >> Using Duke Nukem Forever, which has the mouse jumping bug >> (http://bugs.winehq.org/show_bug.cgi?id=27156) >> >> plain wine: I can move normally, with occasional mouse jumps. >> wine + GrabPointer=N: mouse spins around uncontrollably, a la bug >> http://bugs.winehq.org/show_bug.cgi?id=22282 >> wine configured with --without-xinput2: mouse movement isn't fluid at >> all. Instead, the only way the cursor will move is jumps in a random >> direction (like bug 27156, except there is no normal behavior in >> between the jumps). >> > > Can you check how well it works with wine-1.3.19 - before all xinput2 > related changes went in? Mostly better now, wine-1.3.22-361-g0b2bd0c, though DNF is still broken (http://bugs.winehq.org/show_bug.cgi?id=27572) -- -Austin

June 23, 2011

Re: [PATCH 3/4] shell32: Add IKnownFolder::SetPath() implementation

by Mariusz Pluciński

Please, ignore this single patch. This function should be implemented in other way (using redirection feature, the thing I am working on but did not submit yet). W dniu 23.06.2011 18:50, Mariusz Pluciński pisze: > --- > dlls/shell32/shellpath.c | 32 ++++++++++++++++++++++++++++++-- > dlls/shell32/tests/shellpath.c | 3 +-- > 2 files changed, 31 insertions(+), 4 deletions(-) > > -- Best regards Mariusz Pluciński

June 23, 2011

Re: [2/6] Make ComboBoxEx send CBEN_ENDEDIT when selecting from the dropdown list

by Nikolay Sivov

On Thu, Jun 23, 2011 at 6:57 PM, Jay Yang <jkelleyy(a)gmail.com> wrote: > >> Could you please add a test for that? It's possible to explicitly send >> WM_COMMAND I guess. This could be done as a next patch if you're sure >> it works right. But in general it should come with a fix. > Hi Nikolay, > > Do you have a suggestion on how to simulate selecting an element from the dropdown list? Should I just send a CBN_DROPDOWN message followed by a CBN_CLOSEUP to the ComboBoxEx? > Yes, this could work I think. Sometimes it's a pain to get a reliable test for controls that are made from another ones, because interaction between them differs or creation order differs between native a wine's. If it's not worth it, it's ok to add simple separate test application to reproduce that, link it with a patch or as another message to wine-devel. Just to have more information about what this patch fixes.

June 23, 2011

Re: [4/5] user32: Test that QS_HOTKEY is properly set and cleared.

by Marvin

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=11926 Your paranoid android. === W7PROX64 (64 bit msg) === Timeout

June 23, 2011

Re: [2/6] Make ComboBoxEx send CBEN_ENDEDIT when selecting from the dropdown list

by Nikolay Sivov

On Thu, Jun 23, 2011 at 6:31 PM, Jay Yang <jkelleyy(a)gmail.com> wrote: > Windows send a CBEN_ENDEDIT when selecting from the dropdown menu, and for wine to send that message, the WCBE_ACTEDIT flag needs to be set when the dropdown is created. > --- > dlls/comctl32/comboex.c | 8 ++++---- > 1 files changed, 4 insertions(+), 4 deletions(-) > Hey, Jay. Could you please add a test for that? It's possible to explicitly send WM_COMMAND I guess. This could be done as a next patch if you're sure it works right. But in general it should come with a fix.

June 23, 2011

Re: GSoC-2011: Implement Missing Mesh Functions in Wine’s D3DX9

by Michael Mc Donnell

On Thu, Jun 23, 2011 at 5:50 AM, Dylan Smith <dylan.ah.smith(a)gmail.com> wrote: > On Wed, Jun 22, 2011 at 4:08 PM, Stefan Dösinger <stefandoesinger(a)gmx.at> wrote: >> I'll have to do some more background reading on the kind of mesh data structures used here before I can give qualified comments on your patches, so for now I'll yield to Dylan :-) > > I looked over the latest patches, and the code looks good now, but I > have a couple of nitpicks for the comments. Sounds good :-) > +/* ConvertAdjacencyToPointReps helper function. > + * > + * Goes around the edges of a face and replaces the vertices in any adjacent > > (changed word is in caps, but capitalization not needed) > * Goes around the edges of EACH face... Check. > + * The vertices in a point representation must be ordered sequentially, e.g. > + * index 3 holds the index of the vertex that replaces vertex 3, i.e. if > + * vertex 3 is replaced by vertex 5 then index 3 would contain 5. If no vertex > + * replaces it, then it contains the same number as the index itself, e.g. > + * index 3 would contain 3. */ > > Your example has vertex 5 replacing vertex 3, which doesn't seem > possible if lower vertex indices always replace higher vertex indices. You're right it is the other way around. Thanks for reading it thoroughly. I've change the test a bit. I split up the complex example (mesh6) into two simpler examples. I also got inspired by your ASCII illustrations and have added my own in the comments. The examples have also been updated so that they are easier to draw by hand (except for mesh5).

June 23, 2011

GSoC-2011: Implement Missing Mesh Functions in Wine’s D3DX9

by Michael Mc Donnell

Sorry I made a mistake. On Thu, Jun 23, 2011 at 11:07 AM, Michael Mc Donnell <michael(a)mcdonnell.dk> wrote: > Yes that's the basics of it. It does some other funny things too. For > example it re-orders the indices, so if you have > > 0--1 6 3 > | / / | | \ > | / / | | \ > 2 8--7 5--4 > > That corresponds to: > > index buffer = [ 0,1, 2, 6, 7,8, 3, 4,5] > adjacency = [-1,1,-1, 2,-1,0, -1,-1,1] > point rep = [ 0,1, 2, 1, 4,7, 1, 7,2] Index 5 is not replaced by 7 because the lowest index is always used. So it should have been the other way around: point rep = [ 0,1, 2, 1, 4,5, 1, 5,2] > If the indices had not been re-ordered then the point representation > would have been > [0,1,2, 1,7,2, 1,4,7]. And similarly [0,1,2, 1,5,2, 1,4,5]. I'll put this example into the test.

June 23, 2011

Re: GSoC-2011: Implement Missing Mesh Functions in Wine’s D3DX9

by Michael Mc Donnell

On Thu, Jun 23, 2011 at 3:30 AM, Dylan Smith <dylan.ah.smith(a)gmail.com> wrote: > On Wed, Jun 22, 2011 at 4:08 PM, Stefan Dösinger <stefandoesinger(a)gmx.at> wrote: >> I'll have to do some more background reading on the kind of mesh data structures used here before I can give qualified comments on your patches, so for now I'll yield to Dylan :-) > > I couldn't actually find information on point representative data (aka > PointReps), but a quick search found others wondering what it is. I used chapter 19 of "The Direct3D Graphics Pipeline" by Richard Thomson (his draft book): http://www.xmission.com/~legalize/book/download/19-D3DX%20Mesh%20Objects.pdf > Basically it is an array of vertex indices, one for each vertex, each > with either its own index or the index of an co-located vertex. > e.g. If you have triangles ABC and DEF (as shown below), with two sets > of points co-located, and the vertex buffer stored in alphabetical > order. Then the point_reps array would be [0, 1, 2, 2, 1, 5] to > indicate that vertices B & C can replace E & D respectively. > > A---(B/E) > | / | > | / | > (C/D)--F Yes that's the basics of it. It does some other funny things too. For example it re-orders the indices, so if you have 0--1 6 3 | / / | | \ | / / | | \ 2 8--7 5--4 That corresponds to: index buffer = [ 0,1, 2, 6, 7,8, 3, 4,5] adjacency = [-1,1,-1, 2,-1,0, -1,-1,1] point rep = [ 0,1, 2, 1, 4,7, 1, 7,2] If the indices had not been re-ordered then the point representation would have been [0,1,2, 1,7,2, 1,4,7]. Here is an example that shows how it expands collapsed triangles: 0--1 3 | / / | | / / | 2 4--5 If the triangle 3-4-5 is collapsed by repeating it's first vertex in the index buffer, then it will not be adjacent to 0-1-2: index buffer = [ 0, 1, 2, 3, 3, 3] adjacency = [-1,-1,-1, -1,-1,-1] point rep = [ 0,1, 2, 3, 4,5] The vertices of 3-4-5 are included anyway in the point representation, even though they do not form a real triangle. All these things are tested in mesh6, but I can see now that they probably should be split up into separate test cases.

June 23, 2011

Re: [PATCH 03/10] urlmon: Reimplement CreateURLMonikerEx on top of IUri APIs

by Alexandre Julliard

Jacek Caban <jacek(a)codeweavers.com> writes: > --- > dlls/urlmon/umon.c | 115 > ++++++++++++++++++++++++++++------------------------ > 1 files changed, 62 insertions(+), 53 deletions(-) It doesn't work here: ../../../tools/runtest -q -P wine -M qmgr.dll -T ../../.. -p qmgr_test.exe.so job.c && touch job.ok job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:378: Test failed: Bad state: 4 job.c:386: Test failed: BITS jobs timed out job.c:388: Test failed: IBackgroundCopyJob_Complete job.c:329: Test failed: CreateFile job.c:333: Test failed: ReadFile job.c:338: Test failed: Files differ in length job.c:339: Test failed: Files differ in contents job.c:329: Test failed: CreateFile job.c:333: Test failed: ReadFile job.c:338: Test failed: Files differ in length job.c:339: Test failed: Files differ in contents make: *** [job.ok] Error 39 -- Alexandre Julliard julliard(a)winehq.org

June 23, 2011

Re: GSoC-2011: Implement Missing Mesh Functions in Wine’s D3DX9

by Dylan Smith

On Wed, Jun 22, 2011 at 4:08 PM, Stefan Dösinger <stefandoesinger(a)gmx.at> wrote: > I'll have to do some more background reading on the kind of mesh data structures used here before I can give qualified comments on your patches, so for now I'll yield to Dylan :-) I looked over the latest patches, and the code looks good now, but I have a couple of nitpicks for the comments. +/* ConvertAdjacencyToPointReps helper function. + * + * Goes around the edges of a face and replaces the vertices in any adjacent (changed word is in caps, but capitalization not needed) * Goes around the edges of EACH face... + * The vertices in a point representation must be ordered sequentially, e.g. + * index 3 holds the index of the vertex that replaces vertex 3, i.e. if + * vertex 3 is replaced by vertex 5 then index 3 would contain 5. If no vertex + * replaces it, then it contains the same number as the index itself, e.g. + * index 3 would contain 3. */ Your example has vertex 5 replacing vertex 3, which doesn't seem possible if lower vertex indices always replace higher vertex indices.

June 22, 2011

Re: GSoC-2011: Implement Missing Mesh Functions in Wine’s D3DX9

by Dylan Smith

On Wed, Jun 22, 2011 at 4:08 PM, Stefan Dösinger <stefandoesinger(a)gmx.at> wrote: > I'll have to do some more background reading on the kind of mesh data structures used here before I can give qualified comments on your patches, so for now I'll yield to Dylan :-) I couldn't actually find information on point representative data (aka PointReps), but a quick search found others wondering what it is. Basically it is an array of vertex indices, one for each vertex, each with either its own index or the index of an co-located vertex. e.g. If you have triangles ABC and DEF (as shown below), with two sets of points co-located, and the vertex buffer stored in alphabetical order. Then the point_reps array would be [0, 1, 2, 2, 1, 5] to indicate that vertices B & C can replace E & D respectively. A---(B/E) | / | | / | (C/D)--F The adjacency buffer is documented well enough on MSDN for GenerateAdjacency (http://msdn.microsoft.com/en-us/library/bb205737(v=VS.85).aspx) and seems to be used in this method for determining which vertices are co-located. I hope that helps.

June 22, 2011

Re: Need help at collecting ActionMapping data

by Lucas Zawacki

> Hi, > so you are the One who can finish off the bug 8754?? :-D I hope so. As soon as my first patches get in I hope things should be working with the keyboard, so if you have access to the NFSU games you should test it. > I own force feedback wheel: "Logitech Wingman Formula Force GP" and > did three of this piece of HW on: Awesome! Thank you. > Results are attached. I also have access to some Microsoft Force > Feedback wheel (Sidewinder?), so if interested i could try to make > some log during the weeken. Also Win7 should not be a big problem. > It's not vital but if it's not a problem for you I'd like this data too. Thanks again

June 22, 2011

Re: Need help at collecting ActionMapping data

by wylda＠volny.cz

> You just need to run it, redirect the output to a file... Hi, so you are the One who can finish off the bug 8754?? :-D I own force feedback wheel: "Logitech Wingman Formula Force GP" and did three of this piece of HW on: * WinXP 32 bit SP3 (English) * WinXP 64 bit SP2 (English) * Win Vista Business 32 bit SP2 (Czech :-/ ) Results are attached. I also have access to some Microsoft Force Feedback wheel (Sidewinder?), so if interested i could try to make some log during the weeken. Also Win7 should not be a big problem. Maybe some week translation could come handy in Vista case ;) * "Osa X" ~ X Axis * "Tlačítko" ~ Button * "Kolekce 0 - Pakovy ovladac" ~ Collection 0 - Joystick Regards, W.

June 22, 2011

Re: [PATCH 2/2] d3dx9_36/tests: Added tests for constant table samplers (Try 2)

by Marvin

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=11904 Your paranoid android. === WINEBUILD (build) === Patch failed to apply

June 22, 2011

Re: GSoC-2011: Implement Missing Mesh Functions in Wine’s D3DX9

by Stefan Dösinger

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Am 22.06.2011 um 21:48 schrieb Michael Mc Donnell: >> HeapFree checks for NULL for you, so remove the NULL check before >> calling HeapFree. > > Ok. I see that now in the Wine source. That's a lot nicer than on Windows. Windows also checks for NULL pointers in HeapFree. Otherwise Wine wouldn't do so, and we'd need the NULL checks to be able to run our code on Windows. > My UpdateSemantics patch finally got applied Congratulations! I'll have to do some more background reading on the kind of mesh data structures used here before I can give qualified comments on your patches, so for now I'll yield to Dylan :-) -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.17 (Darwin) iQIcBAEBAgAGBQJOAkvaAAoJEN0/YqbEcdMwxssP/j1BOavQHLCSJdarIwV6B+NL wkIB9W6L6EsvGxr/8we6mTSVAQ3VMBK74F1yB5BBDNls1YXVugDHm7GI5Yrku4lR mE8YhwudJVZ3i3eKeHBvjRIzkGPW+eUpNkkmJQtQVkpU/Lu/igVDB+pxiFSRxhc4 aAA26bIZ3Uer/wm/1YdIi3geJ05CH4uZpDpWaslARURSP4zofsXiGj5m4KI0WCup 7BtlDdwA9SzamxVHRBTXld0Y4vcg80jBOlmvd6soifD+VI4LCXA31xHBCLR//BAK 8mbdDcPF00mJGRwJKXbL/FJunXRocaOSu0ujtc9Tx9BQbtLGAs0C6VWKgfvvh47M jZ0RFc5ACStWQTkSNPrhvycS188oJ/5mW3aISG+AVuTme3H9zSrNccJYnBpjR+iy GRbn2DmmxRziq/XS4GjW7+oCapHwsESoyZNDxO1fqI0k+gODmGWvR41oONN52/wv QbvAxjsLknwMKvX2MTNVAUoeQWsPB9OIV8HxIPjwxNGFsvM8/ETOprHRyZP2dgNc XcnDQM5/5O2amNRCrRIA4tQaE8QbhxQfe0M9fwqf4XatzYx8tzW1kus9lm9HMEp5 j/FZjBnlU+RjSa+u5TDZxeW5nYwPbrw2ygWcTRom4taZydvWHxWBY4t6IkMfSe+g hBPbh6pFtYZwhpFKWD7u =/xC4 -----END PGP SIGNATURE-----

June 22, 2011

Re: GSoC-2011: Implement Missing Mesh Functions in Wine’s D3DX9

by Michael Mc Donnell

On Wed, Jun 22, 2011 at 4:53 PM, Dylan Smith <dylan.ah.smith(a)gmail.com> wrote: > On Wed, Jun 22, 2011 at 10:12 AM, Michael Mc Donnell > <michael(a)mcdonnell.dk> wrote: >> Thanks your comments. I've attached an updated version. > >> + if (indices) HeapFree(GetProcessHeap(), 0, indices); >> + } >> + if (new_indices) HeapFree(GetProcessHeap(), 0, new_indices); > > HeapFree checks for NULL for you, so remove the NULL check before > calling HeapFree. Ok. I see that now in the Wine source. That's a lot nicer than on Windows. >> + struct >> + { >> + ID3DXMesh *meshes[ARRAY_SIZE(vertices)]; >> + const DWORD *indices[ARRAY_SIZE(vertices)]; >> + const DWORD num_vertices[ARRAY_SIZE(vertices)]; >> + const DWORD *adjacencies[ARRAY_SIZE(vertices)]; >> + DWORD *point_reps[ARRAY_SIZE(vertices)]; >> + const DWORD *exp_point_reps[ARRAY_SIZE(vertices)]; >> + const DWORD options[ARRAY_SIZE(vertices)]; >> + } >> + tc = >> + { >> + {NULL, NULL, NULL, NULL, NULL, NULL, NULL}, >> + {indices0, indices1, indices2, indices3, indices4, indices5, indices6, (DWORD*)indices6_16bit}, >> + {num_vertices0, num_vertices1, num_vertices2, num_vertices3, num_vertices4, num_vertices5, num_vertices6, num_vertices6}, >> + {adjacency0, adjacency1, adjacency2, adjacency3, adjacency4, adjacency5, adjacency6, adjacency6}, >> + {point_rep0, point_rep1, point_rep2, point_rep3, point_rep4, point_rep5, point_rep6, point_rep6}, >> + {exp_point_rep0, exp_point_rep1, exp_point_rep2, exp_point_rep3, exp_point_rep4, exp_point_rep5, exp_point_rep6, exp_point_rep6}, >> + {options, options, options, options, options, options, options, options_16bit} >> + }; > > I actually meant something more like: > > --- snip --- > struct { > const DWORD *indices, > const DWORD num_vertices, > const DWORD *adjacencies, > DWORD *point_reps, > const DWORD *exp_point_reps, > const DWORD options, > } tc[] = { > {indices0, num_vertices0, adjacency0, point_rep0, exp_point_rep0, options}, > {indices1, num_vertices1, adjacency1, point_rep1, exp_point_rep1, options}, > ... > } Ah ok :-) That makes more sense. > ... > > for (i = 0; i < ARRAY_SIZE(tc); i++) > { > ... > } > --- snip --- > > I don't think storing an array of meshes is necessary. You can just > release the mesh and set the mesh variable to NULL (as needed to avoid > re-releasing) before creating another one. I've changed it so that it only keeps the first mesh around for the NULL checks. My UpdateSemantics patch finally got applied so I've removed the duplicate functions too.

June 22, 2011

Re: GSoC-2011: Implement Missing Mesh Functions in Wine’s D3DX9

by Dylan Smith

On Wed, Jun 22, 2011 at 10:12 AM, Michael Mc Donnell <michael(a)mcdonnell.dk> wrote: > Thanks your comments. I've attached an updated version. > + if (indices) HeapFree(GetProcessHeap(), 0, indices); > + } > + if (new_indices) HeapFree(GetProcessHeap(), 0, new_indices); HeapFree checks for NULL for you, so remove the NULL check before calling HeapFree. > + struct > + { > + ID3DXMesh *meshes[ARRAY_SIZE(vertices)]; > + const DWORD *indices[ARRAY_SIZE(vertices)]; > + const DWORD num_vertices[ARRAY_SIZE(vertices)]; > + const DWORD *adjacencies[ARRAY_SIZE(vertices)]; > + DWORD *point_reps[ARRAY_SIZE(vertices)]; > + const DWORD *exp_point_reps[ARRAY_SIZE(vertices)]; > + const DWORD options[ARRAY_SIZE(vertices)]; > + } > + tc = > + { > + {NULL, NULL, NULL, NULL, NULL, NULL, NULL}, > + {indices0, indices1, indices2, indices3, indices4, indices5, indices6, (DWORD*)indices6_16bit}, > + {num_vertices0, num_vertices1, num_vertices2, num_vertices3, num_vertices4, num_vertices5, num_vertices6, num_vertices6}, > + {adjacency0, adjacency1, adjacency2, adjacency3, adjacency4, adjacency5, adjacency6, adjacency6}, > + {point_rep0, point_rep1, point_rep2, point_rep3, point_rep4, point_rep5, point_rep6, point_rep6}, > + {exp_point_rep0, exp_point_rep1, exp_point_rep2, exp_point_rep3, exp_point_rep4, exp_point_rep5, exp_point_rep6, exp_point_rep6}, > + {options, options, options, options, options, options, options, options_16bit} > + }; I actually meant something more like: --- snip --- struct { const DWORD *indices, const DWORD num_vertices, const DWORD *adjacencies, DWORD *point_reps, const DWORD *exp_point_reps, const DWORD options, } tc[] = { {indices0, num_vertices0, adjacency0, point_rep0, exp_point_rep0, options}, {indices1, num_vertices1, adjacency1, point_rep1, exp_point_rep1, options}, ... } ... for (i = 0; i < ARRAY_SIZE(tc); i++) { ... } --- snip --- I don't think storing an array of meshes is necessary. You can just release the mesh and set the mesh variable to NULL (as needed to avoid re-releasing) before creating another one.

June 22, 2011

Re: GSoC-2011: Implement Missing Mesh Functions in Wine’s D3DX9

by Michael Mc Donnell

On Tue, Jun 21, 2011 at 5:36 PM, Dylan Smith <dylan.ah.smith(a)gmail.com> wrote: > On Tue, Jun 21, 2011 at 6:32 AM, Michael Mc Donnell > <michael(a)mcdonnell.dk> wrote: >> Here is my test and implementation of ConvertAdjacencyToPointReps. > > >> + hr = iface->lpVtbl->LockIndexBuffer(iface, D3DLOCK_READONLY, (void**)&indices); >> + if (FAILED(hr)) goto cleanup; >> + memcpy(new_indices, indices, This->numvertices * sizeof(*indices)); > > I think you want the number of faces * 3 rather than the number of > vertices. And the size of an index is not constant, it is 32-bit if > This->options has the D3DXMESH_32BIT bit set and 16-bit if the bit > isn't set. Yes you're right, I'll change that. I had completely forgotten about 16-bit indices, so that required a lot more changes and a test. >> +static void test_convert_adjacency_to_point_reps(void) >> +{ > ... >> + /* All mesh data */ >> + struct vertex_pnc *vertices[] = {vertices0, vertices1, vertices2, vertices3, vertices4, vertices5, vertices6}; >> + ID3DXMesh *meshes[ARRAY_SIZE(vertices)]; >> + DWORD *indices[] = {indices0, indices1, indices2, indices3, indices4, indices5, indices6}; >> + DWORD num_vertices[] = {num_vertices0, num_vertices1, num_vertices2, num_vertices3, num_vertices4, num_vertices5, num_vertices6}; >> + DWORD num_faces[] = {num_faces0, num_faces1, num_faces2, num_faces3, num_faces4, num_faces5, num_faces6}; >> + DWORD *adjacencies[] = {adjacency0, adjacency1, adjacency2, adjacency3, adjacency4, adjacency5, adjacency6}; >> + DWORD *point_reps[] = {point_rep0, point_rep1, point_rep2, point_rep3, point_rep4, point_rep5, point_rep6}; >> + DWORD *exp_point_reps[] = {exp_point_rep0, exp_point_rep1, exp_point_rep2, exp_point_rep3, exp_point_rep4, exp_point_rep5, exp_point_rep6}; > > I think it would make sense to put all these arrays of the same length > into a structure (e.g. struct test_case) array. This would make it > clear what needs to be updated to add another test case, and less > error prone than making sure each relevant individual array is > updated. Ok I've tried to make a struct for them. > Also, several of these and similar array seem like they can be made constant. Check, I've const'ified everything I could find. >> +static void test_convert_adjacency_to_point_reps(void) >> +{ > ... >> + for (i = 0; i < ARRAY_SIZE(meshes); i++) >> + { > ... >> + hr = meshes[i]->lpVtbl->ConvertAdjacencyToPointReps(meshes[i], adjacencies[i], point_reps[i]); >> + todo_wine ok(hr == D3D_OK, "ConvertAdjacencyToPointReps failed. " >> + "Got %x expected D3D_OK\n", hr); > > The value i should probably be printed in this error message. > Otherwise, if you see a failure on test.winehq.org that you can't > reproduce, it will be hard to figure out what might have caused the > failure. Yes, good idea. > That's it for now. Thanks your comments. I've attached an updated version.

June 22, 2011

Re: [PATCH 4/6] msvcp90: Added locale class stub

by Alexandre Julliard

Piotr Caban <piotr(a)codeweavers.com> writes: > +/* ?name(a)locale@std@@QBE?AV?$basic_string(a)DU?$char_traits(a)D@std@@V?$allocator(a)D@2@@2(a)XZ */ > +/* ?name(a)locale@std@@QEBA?AV?$basic_string(a)DU?$char_traits(a)D@std@@V?$allocator(a)D@2@@2(a)XZ */ > +DEFINE_THISCALL_WRAPPER(locale_name, 8) > +void __thiscall locale_name(THIS_AND_RET(const locale *this, basic_string_char *ret)) > +{ > + FIXME("(%p %p) stub\n", this, ret); > +} You should declare it correctly for the non-i386 case, the return pointer is a calling convention detail that the compiler should take care of. -- Alexandre Julliard julliard(a)winehq.org

June 22, 2011

Re: [PATCH 1/5] shell32: Add IKnownFolderManager::RegisterFolder implementation (resend)

by Marvin

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=11886 Your paranoid android. === WVISTAADM (32 bit shellpath) === Failure running script in VM: The specified guest user must be logged in interactively to perform this operation

June 22, 2011

[1/2] d3dx9/test: Add UpdateSemantics test. (try 3)

by Michael Mc Donnell

Hi Alexandre My UpdateSemantics patches haven't been applied yet. Stefan Dösinger and Dylan Smith think they look ok. Is there something that looks wrong with them to you, or have you just not have the time to look at them yet? Thanks, Michael Mc Donnell

June 22, 2011