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

dinput8 GSoC wrapup
by Lucas Zawacki Aug. 20, 2011

Aug. 20, 2011

Hey everyone, just sending you this email to give a heads up about the status of my GSoC project. I set out to implement the action mapping features of DirectInput8, what I have now is: * EnumDevicesBySemantics: Fully implemented * SetActionMap: Fully implemented * BuildActionMap: Fully implemented with the exception of loading previous saved configurations. * ConfigureDevices: Functional but still needs more functionalities and some polishment The patches that I accumulated this last 2 weeks and will start to commit as of this next monday are here: * https://gist.github.com/1146808 * https://gist.github.com/1154946 * https://gist.github.com/1159426 * https://gist.github.com/1159498 I'll keep working on the stuff that's missing to have the fully functional action mapping, also I have to commit some documentation and make some cleaning of the code. By the way one nice thing I discovered this last weeks is that we may even be able to use what MSDN calls "vendor defined mappings" and, for instance, have a wheel or other exotic joystick perfectly mapped to the application. What I'll need for this is implementing loading of saved configurations, the device "Vendor ID/Product ID" property and access to the mapping files deployed with a DirectX installation (maybe we could download them with winetricks?). All in all, I'd like to thank you all for helping me trough this project, especially Marcus Meissner and Vitaliy Margolen for reading my patches and giving invaluable advice :)

1 0

random flakiness
by Dan Kegel Aug. 20, 2011

Aug. 20, 2011

When running a test bot, one notices flaky things that otherwise happen too seldom to be noticed. For instance, today, the buildbot crashed once during wineboot: wine: created the configuration directory '/home/dank/wineslave.dir/sandbox/slave/runtests/build/wineprefix' fixme:storage:create_storagefile Storage share mode not implemented. err:mscoree:LoadLibraryShim error reading registry key for installroot ... err:mscoree:LoadLibraryShim error reading registry key for installroot fixme:iphlpapi:NotifyAddrChange (Handle 0x83e8ec, overlapped 0x83e8d0): stub wine: Unhandled page fault on write access to 0x00000000 at address 0x7d8eb051 (thread 0010), starting debugger... err:process:__wine_kernel_init boot event wait timed out When I got home, the crash dialog was still up. ps says the app that crashed was: C:\windows\system32\rundll32.exe setupapi,InstallHinfSection DefaultInstall 128 \\?\unix/home/dank/wineslave.dir/sandbox/slave/runtests/build/tools/wine.inf There have been several other random little failures, e.g. http://bugs.winehq.org/show_bug.cgi?id=28108, http://bugs.winehq.org/show_bug.cgi?id=28133 I guess I'll just grit my teeth and ignore these until they happen more than once, then try to file useful bugs about them. Who knows, maybe it's my system. - Dan

1 0

Re: mshtml: Use wininet to get and parse the proxy
by Dmitry Timoshkov Aug. 20, 2011

Aug. 20, 2011

André Hentschel <nerv(a)dawncrow.de> wrote: > + if(!InternetCrackUrlA(pi->lpszProxy, 0, 0, &UrlComponents)) return; > + heap_free(pi); If InternetCrackUrlA() fails pi is leaked. -- Dmitry.

1 0

Re: [1/2] cmd: fixing an issue with extra whitespaces after cd parameter (retry2)
by Frédéric Delanoy Aug. 20, 2011

Aug. 20, 2011

On Fri, Aug 19, 2011 at 02:46, Nowres Rafid <nowres.rafed(a)gmail.com> wrote: > else { + pos = string; /* Remove any double quotes, which may be in the middle, eg. cd "C:\Program Files"\Microsoft is ok */ - pos = string; > That's a gratuitous change while (*command) { if (*command != '"') *pos++ = *command; command++; } *pos = 0x00; + /* Remove any whitespace at the end of command + This was because "cd .. " wasn't working */ + pos--; + while ( *pos == ' ' || *pos == '\t' ) + *pos-- = 0x00; + You don't need to alter the string more than necessary. I would do sthg like while (*command) { if (*command != '"') *pos++ = *command; command++; } + pos--; + while (*pos == ' ' || *pos == '\t') pos--; *pos = 0x00;

1 0

Re: [2/2] cmd/tests: fixing an issue with extra whitespaces after cd parameter (retry)
by Frédéric Delanoy Aug. 20, 2011

Aug. 20, 2011

You should use @space@ elements to make the tests clearer. Also, you shouldn't use (retry) in your patch mail subject, but rather (try N)

1 0

What gstreamer plugins are actually needed?
by Scott Ritchie Aug. 19, 2011

Aug. 19, 2011

I'm trying to cleanup a bit of a packaging mess whereby Wine currently pulls in 250 or so gstreamer plugins. Which ones are actually useful for applications? Do we even know? Thanks, Scott Ritchie

2 1

Buildbot status
by Dan Kegel Aug. 19, 2011

Aug. 19, 2011

http://buildbot.kegel.com/ caught a couple of real problems today. Remember the original patchwatcher? It died because only one person knew how to run it, and it was too hard to set up. I'm trying to not make that mistake this time. To that end, I've made the process easier, and would like as many people as possible to try running buildbot slaves (and, at least temporarily, masters!) so the knowledge is spread around. It's probably ready for other people to try running one. If you're interested, see http://code.google.com/p/winezeug/source/browse/trunk/buildbot/readme-wine.… Changes since last post - made cleaner script and doc for adding a new slave - rewrote patch series recognizer (old one was broken and opaque) to-do list: - make it start automatically after reboot - sandbox it - turn on build result emails - add a bit of parallelism into the test runner, so e.g. tests that don't need the display or network run in parallel with those that do, probably in a separate wineprefix - add valgrind slave - add windows slave (like testbot, but with real graphics and sound) - add x86_64 slave - move master to winehq.org once it seems ready

3 5

re: regedit: adding confirmation message before importing from file
by Dan Kegel Aug. 19, 2011

Aug. 19, 2011

Hi Nowres, did you run the regedit tests, cd programs/regedit/tests make test after that change? Here it fails with regedit.c:384: Test failed: regedit not available, skipping regedit tests ( log at http://buildbot.kegel.com/builders/runtests/builds/2 ) You might need to fully implement the /S switch for regedit and use it from the tests before you can add the confirmation dialog to regedit. - Dan

2 3

Only use D3DLOCK_DISCARD if dynamic memory (was Re: d3dx9: CloneMesh test and improvements)
by Michael Mc Donnell Aug. 19, 2011

Aug. 19, 2011

On Fri, Aug 19, 2011 at 2:20 PM, Stefan Dösinger <stefandoesinger(a)gmx.at> wrote: > Am 19.08.2011 um 14:04 schrieb Michael Mc Donnell: >> Ok I see that in dlls/d3d9/tests/buffers.c. So it will fail to lock on >> Win7 if D3DLOCK_DISCARD is specified and the vertex buffer wasn't >> created with D3DUSAGE_DYNAMIC? > Yes, unless Microsoft changed this again. > >> Am I correct in D3DLOCK_DISCARD makes a copy of the vertex buffer so >> that the video card can keep reading the old one, and the old vertex >> buffer memory is released after the video card releases its lock [1]? > Sort of. It doesn't copy the old data, it just gives you a fresh, uninitialized block of memory. So if you read from the pointer you get from a DISCARD lock you'll get garbage. > >> In that case it does not make sense to use D3DLOCK_DISCARD at all in >> CloneMesh, as the video card has not begun to use the new vertex >> buffer, and it will just add memory overhead. The easiest fix would be >> to just pass 0 instead of D3DLOCK_DISCARD. I've done that in the >> updated patch. > Right, the buffer is newly created in CloneMesh. Just not using DISCARD is the best fix then. > >>> I noticed a few more D3DLOCK_DISCARD locks in the mesh.c code. >> >> I guess the other methods should check the mesh options for >> D3DXMESH_DYNAMIC, D3DXMESH_VB_DYNAMIC, and D3DXMESH_IB_DYNAMIC [2] >> before using D3DLOCK_DISCARD? I'll write patches for those. > Most likely yes. I haven't looked at the details. If the buffer is new in those situations as well just don't pass any flags, otherwise pass DISCARD if you lock a dynamic buffer and are going to rewrite the entire buffer contents. All of the places with D3DLOCK_DISCARD in mesh.c were working on newly created meshes or vertex buffers, so they should all pass 0. I've attached patches for it.

1 0

re: shlwapi: Put correct return values in SetFromTimeInterval
by Dan Kegel Aug. 19, 2011

Aug. 19, 2011

(The buildbot worked its way back to the day after Alexandre went on vacation. Even though the patches are old, git hasn't changed, so they ought to apply still.) It didn't apply here: patching file dlls/shlwapi/string.c patching file dlls/shlwapi/tests/string.c Hunk #1 FAILED at 622. Hunk #2 FAILED at 639. Hunk #3 FAILED at 649. 3 out of 3 hunks FAILED -- saving rejects to file dlls/shlwapi/tests/string.c.rej Log at http://buildbot.kegel.com/builders/runtests/builds/80 for the next day or so.

1 0

Re: About cmd patches
by Nowres Rafid Aug. 19, 2011

Aug. 19, 2011

On 19/08/2011 03:14, Dan Kegel wrote: > Hi Nowres, > it's awesome that you're digging into cmd like this! > > The rule with Wine is that > Each patch you submit shall pass tests. > > That means that your patch series > [1/2] cmd: fixing an error with redirection operators > [2/2] cmd/tests: fixing an error with redirection operators (updating > test results) > > is incorrect, since tests marked todo_wine will fail after the first patch. > > Better to send things like that as a single merged patch in the future. > > Thanks, > Dan Ok Dan, i queued these patch in my email client, but when I sent them they were sent in bad order. Nowres,

2 1

d3dx9: CloneMesh test and improvements
by Michael Mc Donnell Aug. 19, 2011

Aug. 19, 2011

Hi I've been working on a test and improvements for CloneMesh as part of GSoC 2011. I would be grateful if anyone could comment on my work. There are five test cases in the attached patch: 0. Basic mesh cloning. Declaration has position and normal, and the new declaration is the same as the original. 1. Same as test 0, but with 16-bit indices. 2. Shows that the vertex buffer can be widened so that it has room for a new vertex component. The declaration has position and normal, and the new declaration has texture coordinates after position and normal. 3. Same as test 2, but the new declaration has the texture coordinates in between position and normal. 4. Shows that a vertex component can be converted into another type if the new declaration specifies another type. The declaration has position and normal. The normal is a FLOAT2 and in the new declaration it is instead FLOAT16_2. Test 0 and 1 tests what is already known to work in the current implementation. Test 2, 3 and 4 do not work in the current implementation because the original and new declarations are not the same. I've also improved the current CloneMesh implementation so that it also passes test 2, 3, and 4. The patch is not complete because it does not support converting all the possible types. I expect to implement conversion for the remaining types during this week. Again any comments or ideas are welcome. Thanks, Michael Mc Donnell

5 15

re: mscoree: COM cleanup for the ICLRRuntimeInfo iface.
by Dan Kegel Aug. 19, 2011

Aug. 19, 2011

Seems to cause a test failure here? make[1]: Entering directory `/home/dank/wineslave.dir/sandbox/slave/runtests/build/dlls/mscoree/tests' ../../../tools/runtest -q -P wine -M mscoree.dll -T ../../.. -p mscoree_test.exe.so metahost.c && touch metahost.ok metahost.c:135: Test failed: got unexpected version L"v1.1.4322" fixme:mscoree:CLRMetaHost_GetRuntime Unrecognized version L"v2.0.0" Command exited with non-zero status 1 0.04user 0.03system 0:00.09elapsed 82%CPU (0avgtext+0avgdata 27104maxresident)k 0inputs+0outputs (0major+5905minor)pagefaults 0swaps make[1]: *** [metahost.ok] Error 1 ../../../tools/runtest -q -P wine -M mscoree.dll -T ../../.. -p mscoree_test.exe.so mscoree.c && touch mscoree.ok mscoree.c:140: Test failed: version is L"v1.1.4322" , expected L"v2.0.50727" mscoree.c:145: Test failed: version is L"v1.1.4322" , expected L"v2.0.50727" fixme:mscoree:CreateConfigStream ((null), 0x32fbb4): stub fixme:mscoree:CreateConfigStream (L"Z:\\home\\dank\\wineslave.dir\\sandbox\\slave\\runtests\\build\\dlls\\mscoree\\tests\\conf.xml", (nil)): stub fixme:mscoree:CreateConfigStream ((null), (nil)): stub fixme:mscoree:CreateConfigStream (L"nonexist.xml", 0x32fbb4): stub fixme:mscoree:CreateConfigStream (L"Z:\\home\\dank\\wineslave.dir\\sandbox\\slave\\runtests\\build\\dlls\\mscoree\\tests\\conf.xml", 0x32fbb4): stub Command exited with non-zero status 2 0.05user 0.02system 0:00.09elapsed 81%CPU (0avgtext+0avgdata 27296maxresident)k 0inputs+8outputs (0major+5969minor)pagefaults 0swaps make[1]: *** [mscoree.ok] Error 2 make[1]: Target `test' not remade because of errors. (Log at http://buildbot.kegel.com:8010/builders/runtests/builds/32 for a day or so.)

2 2

Re: winemenubuilder: appbundle patch 20110618
by Per Johansson Aug. 19, 2011

Aug. 19, 2011

Hi Steven, list, I tried both AppleScript and shell script versions and AppleScript doesn't really add anything, so it's probably better to go with shell script and generate the bundle manually. I'll see if I can help smoothing it out. -- Pelle Johansson 17 aug 2011 kl. 00.06 skrev Steven Edwards: > You might be interested in this. It allows icons and localized > resources. I've just not gotten around to doing all the polish. > > > ---------- Forwarded message ---------- > From: Steven Edwards <winehacker(a)gmail.com> > Date: Sat, Jun 18, 2011 at 5:43 AM > Subject: winemenubuilder: appbundle patch 20110618 > To: Wine Developers List <wine-devel(a)winehq.com> > > > Hi, > I've finally gotten back around to hacking the *.app bundle support in > winemenubuilder and come close to having something that mostly works. > There are still things that need to be cleaned up and further > abstracted, and a bit more logic to work out. I wanted to go ahead and > send this out in case anyone else wants to play around with it, or if > for some reason I am unable to finish hackin on it. > > Some notes: > > - The first icon/bundle always seems to get the wrong icon, and uses > the default wine icon. I need to take some time to understand the code > a bit better and see why this is. > > - Bundles are created in $HOME/Applications/wine and icon's are > generated using the icns code. Support for building bundles for > shortcuts on $HOME/Desktop has not been hacked on yet. > > - There are still plenty of bugs to resolve such as making sure we are > actually using the correct icon, and properly setting various values > in the app *.plist. There are a couple of alternative ideas I looked > in to, such as doing some sort of spawn/fork to invoke > /usr/libexec/PlistBuddy rather than using the CoreFoundation API, but > I worry about creating race conditions and other performance problems > when installing software that creates many icons, such as Microsoft > Office. > > - I've tried to make this un-invasive to the existing xdg code path > (as one can run a FreeDesktop on Darwin/OS X) but have not had a > chance to actually test supporting both menu systems at the same time. > > I welcome your feedback. > Thanks > -- > Steven Edwards > > "There is one thing stronger than all the armies in the world, and > that is an idea whose time has come." - Victor Hugo > > > > -- > Steven Edwards > > "There is one thing stronger than all the armies in the world, and > that is an idea whose time has come." - Victor Hugo > <appbundles_20110617.diff>

1 0

re: cmd/tests: fixing an issue with extra whitespaces after cd parameter
by Dan Kegel Aug. 18, 2011

Aug. 18, 2011

That patch will fail unless your earlier patch is applied, right? If that's the case, you have to submit the two patches together, either as a single patch, or as a two patch series ( see http://wiki.winehq.org/SubmittingPatches for more info, in particular, how to format a two patch series ).

1 0

re: Subject: winemenubuilder: Use start.exe to open URLs to respect URL handlers in registry.
by Dan Kegel Aug. 18, 2011

Aug. 18, 2011

It doesn't apply here, looks like whitespace was corrupted. Try attaching the patch instead of sending it inline? patching file programs/winemenubuilder/winemenubuilder.c Hunk #1 FAILED at 2962. Hunk #2 FAILED at 3004. Hunk #3 FAILED at 3063. 3 out of 3 hunks FAILED -- saving rejects to file programs/winemenubuilder/winemenubuilder.c.rej (Log at http://buildbot.kegel.com:8010/builders/runtests/builds/22/steps/git/logs/s… for a few days.)

1 0

re: cmd: fixing an issue with extra whitespaces after cd parameter (retry)
by Dan Kegel Aug. 18, 2011

Aug. 18, 2011

Frédéric Delanoy wote: > Nowres Rafid wrote: >> how can I generate the expected file (.exp)? > It already exists in the source tree I think Nowres meant, how can he update the .exp file to match the expected output of his new test. The answer is: run the new test, save its output, and add it to the .exp file by hand. It's kind of a pain.

2 2

Re: xlive: stub some functions (2/3)
by Austin English Aug. 18, 2011

Aug. 18, 2011

On Wed, Aug 17, 2011 at 12:33, Giovanni Ongaro <giovanni.nicola(a)ticino.com> wrote: > I am working oon a xlive patch can you tell me how to finf dll ordinal > functions by using winedump i am tryiing specifically to find how much > parameters xlive 5282 has > > Can you help me? >From the patch: +/* This DLL is completely undocumented. DLL ordinal functions were found using winedump + and number of parameters determined using Python ctypes. See http://docs.python.org/library/ctypes.html */ -- -Austin

1 0

buildbot status
by Dan Kegel Aug. 17, 2011

Aug. 17, 2011

I brought up the proof-of-concept buildbot-based patchwatcher on two old PCs here; its status page is http://buildbot.kegel.com/ It has caught one real problem so far this morning. to-do list: - make cleaner script and doc for adding a new slave - make it start automatically after reboot - sandbox it - turn on build result emails - add a bit of parallelism into the test runner, so e.g. tests that don't need the display or network run in parallel with those that do, probably in a separate wineprefix - add valgrind slave - add windows slave (like testbot, but with real graphics and sound) - add x86_64 slave - move master to winehq.org once it seems ready Here are the tests I had to blacklist so far: # http://bugs.winehq.org/show_bug.cgi?id=12053 touch dlls/user32/tests/msg.ok touch dlls/user32/tests/win.ok touch dlls/user32/tests/input.ok # http://bugs.winehq.org/show_bug.cgi?id=28038 touch dlls/wininet/tests/urlcache.ok # http://bugs.winehq.org/show_bug.cgi?id=28109 touch dlls/winmm/tests/capture.ok # http://bugs.winehq.org/show_bug.cgi?id=28048 touch dlls/winmm/tests/wave.ok # Blacklist until http://www.winehq.org/pipermail/wine-patches/2011-August/105358.html in touch dlls/winhttp/tests/winhttp.ok # http://bugs.winehq.org/show_bug.cgi?id=28102 touch dlls/ws2_32/tests/sock.ok # http://bugs.winehq.org/show_bug.cgi?id=28108 touch dlls/urlmon/tests/url.ok

1 0

Re: cmd: fixing an issue with extra whitespaces after cd parameter (retry)
by Frédéric Delanoy Aug. 17, 2011

Aug. 17, 2011

2011/8/17 Nowres Rafid <nowres.rafed(a)gmail.com>: > On 17/08/2011 14:02, Frédéric Delanoy wrote: >> >> On Fri, Aug 12, 2011 at 19:14, Nowres Rafid<nowres.rafed(a)gmail.com> >> wrote: >>> >>> fixing cd ..(a)space@ >>> >>> please ignore the previous patch. >> >> You didn't include any test in your patch. >> All patches including a fix must have tests confirming the bug is fixed > > ok, > > how can I generate the expected file (.exp)? It already exists in the source tree

1 0

Re: cmd: fixing an issue with extra whitespaces after cd parameter (retry)
by Frédéric Delanoy Aug. 17, 2011

Aug. 17, 2011

On Fri, Aug 12, 2011 at 19:14, Nowres Rafid <nowres.rafed(a)gmail.com> wrote: > fixing cd ..(a)space@ > > please ignore the previous patch. You didn't include any test in your patch. All patches including a fix must have tests confirming the bug is fixed

1 0

Vista/w2k8/w7 users, please test mmdevapi capture
by Joerg-Cyril.Hoehle＠t-systems.com Aug. 17, 2011

Aug. 17, 2011

Dear users of machines with Vista/w2k8/w7, Unfortunately, testbot has no mmdevapi-capture enabled machine. If you have a capture-enabled sound card, could you please run the attached patch and report results to me? How you can tell if capture/recording works on your machine: Run the regular mmdevapi:capture and render tests as well as winmm:* (mci, wave etc.) and see if and why they skip tests. You'll find a binary mmdevapi_test.exe (and test64.exe) on testbot: https://testbot.winehq.org/JobDetails.pl?Key=13426 Testbot's machines say: capture.c:724: Tests skipped: No sound card available capture: 1 tests executed (0 marked as todo, 0 failures), 1 skipped. 1 test is too little and shows capture was skipped. Background: I wrote a patch to fix the Get/ReleaseBuffer protocol in the renderer. Looking at the capture code, I suppose a similar patch could apply, but I've no machine at all to test it! ("Protocol" means the order in which methods can be invoked and implies testing the error codes and effects). Thank you for testing, Jörg Höhle

3 3

Re: regedit: adding confirmation message before importing from file
by Andrew Eikum Aug. 17, 2011

Aug. 17, 2011

On Wed, Aug 17, 2011 at 12:24:18PM +0000, Nowres Rafid wrote: > - reg_file = fopen(realname, "r"); > + > + reg_file = fopen(realname, "r"); The file uses spaces for indenting. Please be consistent and don't use tabs. This is true elsewhere, too: > STRINGTABLE > BEGIN > + IDS_IMPORT_BOX_TEXT "Are you sure you want to add file informations to the registry?" > + IDS_IMPORT_SUCCESS_BOX_TEXT "Values successfuly added to the registry" > + IDS_IMPORT_FAIL_BOX_TEXT "Failed to add values to the registry" > +END Andrew

1 0

Listed package maintainers
by Ben Klein Aug. 17, 2011

Aug. 17, 2011

Hello everyone, For those of you who remember all the trouble I've had with my health, I'm on the mend now. I'm adjusting to some new medication that's doing wonders, and also adjusting to the new lifestyle they allow me to lead. I've just had a look at http://www.winehq.org/download, as I'm well out-of-date with Wine, and I see Maurilio and I are still listed as being the maintainers for the Debian packages, even though Kai Wasserbäch has taken over that duty (and thank you Kai for doing so!) I also notice the pages have been renamed (so instead of deb = ubuntu, ubuntu and debian are named correctly and unambiguously). I would like to voice my approval as former maintainer of the Debian packages, as token as that may seem at this stage. Kind regards, Ben

3 4

Re: dpnet: add missing IDirectPlay8Peer methods
by Michael Stefaniuc Aug. 17, 2011

Aug. 17, 2011

Luis, Louis Lenders wrote: > This is part of bug http://bugs.winehq.org/show_bug.cgi?id=28008 a line length of 400+ chars kinda stretches what is acceptable. Please try to stick to a length of 100 in dpnet; that's the Wine preferred line length and also what I used during the COM cleanup in dpnet. Please also stick with one form of placing the spaces, either "TYPE * const var" or "TYPE *const var" but not both in the same patch. I don't remember which form I have used. And please try to avoid PVOID and LPVOID, even if MSDN documents the methods that way. Those are just gratuitous types that add no value. thanks bye michael

1 0