Re: wineps.drv: Remove variables scale_xx, scale_xy, scale_yx, and scale_yy and two dozen lines of dead code from append_complex_glyph.
On Sun, 9 May 2010, Nikolay Sivov wrote: I don't think it's dead. Variable ptr is used in a cycle. You are amazing. And right. I just submitted an updated patch. Gerald
Re: gdi32/tests: Remove variable oldPen which is not really used from test_widenpath.
On Sun, 9 May 2010, Vitaliy Margolen wrote: On 05/07/2010 01:06 PM, Gerald Pfeifer wrote: -oldPen = SelectObject(hdc, greenPen); +SelectObject(hdc, greenPen); If it's not used it's a bug. Everything should be reset to original state to prevent influence on following tests. There are number of such places in Wine's test suite that makes it impossible to isolate the problem. Hmm, the function test_widenpath starts with HDC hdc = GetDC(0) and ends with ReleaseDC(0, hdc). Do you suggest to SelectObject(hdc,oldPen) before the latter? Isn't that a no-op? There are two instances in that function where a pen is selected. Do you suggest to restore to the old pen before the second? Looking at the test, I'm not sure how relevant that would be. (Alexandre already committed my original patch, but I don't want to just ignore your feedback. It's well possible I may be missing something.) Gerald
Re: Sudden increase of Valgrind errors
On 02/28/2010 03:49 PM, Dan Kegel wrote: Two things: I moved to karmic (and now more modules are enabled, since it comes with new-enough libraries), and I updated valgrind. I haven't had time to dig in to the differences. The suppressions file also needs updating for karmic. Hi Dan, What's the status on the Valgrind reporting you were doing (automatically)? The last one is from March 18th if I'm correct. I'm not sure if fixing Valgrind errors would be accepted in the code freeze period but I still find it useful to have such reporting. The same is true for Coverity. There are quite a lot false positives and several errors can be ignored but on the other hand loads of (potential) bugs were fixed because of these reports. -- Cheers, Paul.
Re: shdocvw: Added IDataObject stub implementation.
Ilya Shpigor shpi...@etersoft.ru writes: --- dlls/shdocvw/Makefile.in |1 + dlls/shdocvw/dataobject.c | 127 + dlls/shdocvw/shdocvw.h|3 + dlls/shdocvw/webbrowser.c |4 ++ 4 files changed, 135 insertions(+), 0 deletions(-) create mode 100644 dlls/shdocvw/dataobject.c Please don't create a new file for every interface. There are enough existing files where this can go. Same for IOleLink. -- Alexandre Julliard julli...@winehq.org
Release plans
Hi, The 64-bit support is now more or less complete I hope I can finish my MCI parser patches in time. Without them, every 64bit app using MCI string commands is likely to crash (OTOH MCI commands work (those using the MCI_*_PARAMS structures)). What can Mac users expect from this release? Yesterday I installed an app with wine-1.1.44 on Mac OS (instead of moving it from a Linux install). It was disappointing from a user POV: - Wine created .desktop files that work on Linux but don't make sense on Mac OS. Here I've been thinking about a simple patch that would instead generate a .command file like I've described in the FAQ. OTOH, Steven Edwards IIRC once had a much more elaborate patch about better Mac OS integration, rejected last year. - In the hidden directory ~/.local/share/icons/ it created .xpm files that the Finder does not display. .png would be displayed. I have no idea how the other packages built atop Wine behave on MacOS behave: - Kronenberg's WineBottler - doh's WineSkin - Macports build - CodeWeaver's CrossOver4Mac I assume they create nice icons that the user can click after an install. Regarding the former 2 packages, I've always been wondering why there's some sort of split in the Wine+Mac user community. Obviously the stock Wine fails to meet the Mac user's expectations, such that several people start and write something better integrated with the GUI -- but not integrated into the Wine source. And I didn't write trivial Mac patches either, e.g. to have wineprefixcreate symlink c:\users\xyz\ Desktop + Videos + Documents + Music to /Users/xyz/Desktop/ etc. This happens on Linux, not on MacOS. That's another (example of a) missing element. (Why didn't I write it? Because I was unsure where to put the #ifdef) Regards, Jörg Höhle
Re: gdi32: use usp10 to optionally generate glyphs for bidi strings
Hi Dmitry, I know this is not official MSDN documentation but this appears to disagree with you. http://www.catch22.net/tuts/neatpad/11 It also makes sense to have all the complex script processing logic in one place instead of spreading it out and duplicating it. Why do you say that Windows gdi32 does not use usp10? I do not see a direct dependency but I have not traced inside to see if it is doing LoadProcAddress or the like. -aric Dmitry Timoshkov wrote: Aric Stewart a...@codeweavers.com wrote: allows us to be able to make use of the mirroring code and eventually the shaping code when it is in place. This should be done the other way around. gdi32 should not depend on a high level dll (that creates circular dependencies), gdi32 in Windows doesn't use usp10 either.
Re: Sudden increase of Valgrind errors
On Fri, May 14, 2010 at 2:22 AM, Paul Vriens paul.vriens.w...@gmail.com wrote: What's the status on the Valgrind reporting you were doing (automatically)? The last one is from March 18th if I'm correct. It got lost in the shuffle (sore wrists, busy day job, etc.) but I should be able to start it up again. (Once I have the 3d performance reporting bot happy, anyway.) I'm not sure if fixing Valgrind errors would be accepted in the code freeze period but I still find it useful to have such reporting. At the very least, catching and fixing valgrind regressions introduced during the code freeze is absolutely a good idea. And it'd probably be a good idea to look at all the tests I don't currently run under valgrind because they make it explode. - Dan
Re: winedump: Print the network share name in .lnk files.
@@ -245,6 +254,18 @@ static int dump_location(void) } printf(\n); +/* dump information about the network volume the link points to */ +printf(Network volume ofs= %08x , loc-dwNetworkVolTableOfs); +if (loc-dwNetworkVolTableOfs (loc-dwNetworkVolTableOfsloc-dwTotalSize)) me wonders if if (loc-dwNetworkVolTableOfs (loc-dwNetworkVolTableOfs + sizeof(NETWORK_VOLUME_INFO)loc-dwTotalSize)) wouldn't be better (same also would apply for test in old code above for dwVolTableOfs -- -- Eric Pouech
Re: gdi32: use usp10 to optionally generate glyphs for bidi strings
Aric Stewart a...@codeweavers.com wrote: I know this is not official MSDN documentation but this appears to disagree with you. http://www.catch22.net/tuts/neatpad/11 It also makes sense to have all the complex script processing logic in one place instead of spreading it out and duplicating it. Bidi and reordering were supported by gdi32 before Uniscribe has been introduced. Accridong to http://msdn.microsoft.com/en-us/goglobal/bb688137.aspx uniscribe is used by lpk.dll (language packs): ExtTextOut can be used to lay out multilingual Unicode text including complex scripts. There is no need for you to do anything other than call ExtTextOut; it handles everything for you. Why do you say that Windows gdi32 does not use usp10? I do not see a direct dependency but I have not traced inside to see if it is doing LoadProcAddress or the like. Inspecting strings in gdi32.dll should be enough. -- Dmitry.
Re: gdi32: use usp10 to optionally generate glyphs for bidi strings
ok, so the LPK calls uniscribe. Do you feel we should implement the LPK style of interface to gdi32? It seems needlessly cumbersome to me. The LPK.dll interfaces seem to be undocumented but based on names should not be to difficult to figure out. -aric Dmitry Timoshkov wrote: Aric Stewart a...@codeweavers.com wrote: I know this is not official MSDN documentation but this appears to disagree with you. http://www.catch22.net/tuts/neatpad/11 It also makes sense to have all the complex script processing logic in one place instead of spreading it out and duplicating it. Bidi and reordering were supported by gdi32 before Uniscribe has been introduced. Accridong to http://msdn.microsoft.com/en-us/goglobal/bb688137.aspx uniscribe is used by lpk.dll (language packs): ExtTextOut can be used to lay out multilingual Unicode text including complex scripts. There is no need for you to do anything other than call ExtTextOut; it handles everything for you. Why do you say that Windows gdi32 does not use usp10? I do not see a direct dependency but I have not traced inside to see if it is doing LoadProcAddress or the like. Inspecting strings in gdi32.dll should be enough.
Re: gdi32: use usp10 to optionally generate glyphs for bidi strings
Dmitry Timoshkov dmi...@codeweavers.com writes: Aric Stewart a...@codeweavers.com wrote: I know this is not official MSDN documentation but this appears to disagree with you. http://www.catch22.net/tuts/neatpad/11 It also makes sense to have all the complex script processing logic in one place instead of spreading it out and duplicating it. Bidi and reordering were supported by gdi32 before Uniscribe has been introduced. Accridong to http://msdn.microsoft.com/en-us/goglobal/bb688137.aspx uniscribe is used by lpk.dll (language packs): ExtTextOut can be used to lay out multilingual Unicode text including complex scripts. There is no need for you to do anything other than call ExtTextOut; it handles everything for you. Which of course demonstrates that gdi32 calls usp10 on native too. Maybe it does it indirectly through lpk.dll, but the end result is the same, you have a dependency on usp10. -- Alexandre Julliard julli...@winehq.org
Re: Release plans
Hey Brian, Jeremy - do you have a copy of the real press release we did for 1.0? I dug around looking for it and couldn't find it. Looks like we never properly posted it on WineHQ. It did get picked up by quite a few news sites, but Google isn't finding it. Scott / Edward - when 1.0 came out, Jeremy had the PR firm that writes copy for Codeweavers put together a press release for Wine. It was pretty good and fairly polished. I'm not sure if having them do it again would be an option. Also, I think Wine 1.0 was coordinated with a release of Crossover 7.0. I can't find anything on that release, but we're certainly happy to put together another one for Wine 1.2. I've CC'd Jon Parshall, as he's the guy that'll get to do it. Cheers, Jeremy
Re: [PATCH 1/3] wined3d: fix texture conversion flag bug
The patch looks like it makes sense at first sight (though the double conversion check is less than ideal), but you really want something more descriptive for the Changelog.
Re: Release plans
On Sat, May 15, 2010 at 1:11 AM, Jeremy White jwh...@codeweavers.com wrote: I can't find anything on that release, but we're certainly happy to put together another one for Wine 1.2. I've CC'd Jon Parshall, as he's the guy that'll get to do it. Could you link us to a copy of the 1.0 one? Some one professional doing the release would be really cool. If that is sorted I really want to do a short showcase of applications that do run well now under Wine (and maybe Crossover? ie Chrome) with a blurb and screen shot for them. A static page that can be linked to for people who want to quickly check out the progress instead of trudging through the appdb.
Re: Release plans
Edward Savage wrote: On Sat, May 15, 2010 at 1:11 AM, Jeremy White jwh...@codeweavers.com wrote: I can't find anything on that release, but we're certainly happy to put together another one for Wine 1.2. I've CC'd Jon Parshall, as he's the guy that'll get to do it. Could you link us to a copy of the 1.0 one? Um, no; that's what I meant when I said I couldn't find anything on it grin. In fact, our PR firm is questioning our sanity - are we really sure we did a release? I'm not sure a release is all that advantageous; it goes over the wire to overwhelmed newsprint reporters who usually ignore it. An announcement is quite likely to be picked up on places like Slashdot and Digg, and that then tends to spur internet buzz far and wide. So...an announcement may be all we really need. Cheers, Jeremy
Re: [2/2] dlls/crypt32: unit test PFXImportCertStore() (2nd try)
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=2067 Your paranoid android. === W98SE (32 bit decode) === decode.c:236: Test failed: succeeded in importing garbage certificate {0x57} decode.c:249: Test failed: could not import the certificate store with a blank password {0x57} === WNT4WSSP6 (32 bit decode) === decode.c:236: Test failed: succeeded in importing garbage certificate {0x57} decode.c:249: Test failed: could not import the certificate store with a blank password {0x57}
Re: More benchmarks of 3d performance comparing windows and linux
Am Donnerstag 13 Mai 2010 03:10:10 schrieb Dan Kegel: I hear there's a +timedemoquit command Can you send me the timedemo files you use? My timedemo is this one here: http://stud4.tuwien.ac.at/~e0526822/mydemo.dem.bz2 It is quite old, but still works. It is a rather long one, but I recorded it when Wine ran this game at ~25 fps. Nowadays it runs at 90-100 fps, so the demo runs rather fast.
Re: More benchmarks of 3d performance comparing windows and linux
On Fri, May 14, 2010 at 10:19 AM, Stefan Dösinger stefandoesin...@gmx.at wrote: My timedemo is this one here: http://stud4.tuwien.ac.at/~e0526822/mydemo.dem.bz2 It is quite old, but still works. It is a rather long one, but I recorded it when Wine ran this game at ~25 fps. Nowadays it runs at 90-100 fps, so the demo runs rather fast. It ran nicely (if gruesomely). But how do you get framerates out of this (on Windows)? http://www.digital-daily.com/video/hl2-benchmarking/ claims it shows them in the console, and saves them to the file hl2hl2source.csv, but neither of those seemed to happen. Here's what I did: cd ~/.wine/drive_c/Program Files/Steam cp ~/mydemo.dem config wine steam.exe -login myuser mypass -applaunch 220 -novid -console -window -w 1024 -h 768 -dxlevel 90 +timedemoquit mydemo.dem I also ran it with WINEDEBUG=+file and +timedemo, but there was no .csv opened ever :-( - Dan
Re: More benchmarks of 3d performance comparing windows and linux
On Fri, May 14, 2010 at 11:27 AM, Dan Kegel d...@kegel.com wrote: http://stud4.tuwien.ac.at/~e0526822/mydemo.dem.bz2 It ran nicely (if gruesomely). But how do you get framerates out of this (on Windows)? http://www.digital-daily.com/video/hl2-benchmarking/ claims it shows them in the console, and saves them to the file hl2hl2source.csv, http://www.techenclave.com/gaming/half-life-2-tweaks-cheats-mods-10247.html is more specific, it says timedemo [demoname] - Plays the specified demo and reports performance information upon completion, including frames played, time taken, average FPS and FPS variability. Also records the information in a file called sourcebench.csv in your \Program Files\Valve\Steam\SteamApps\[username]\half-life 2\hl2\ directory. I see that directory, but there's no new file in it :-( - Dan
Re: More benchmarks of 3d performance comparing windows and linux
On Fri, May 14, 2010 at 11:32 AM, Dan Kegel d...@kegel.com wrote: On Fri, May 14, 2010 at 11:27 AM, Dan Kegel d...@kegel.com wrote: http://stud4.tuwien.ac.at/~e0526822/mydemo.dem.bz2 It ran nicely (if gruesomely). But how do you get framerates out of this (on Windows)? http://www.digital-daily.com/video/hl2-benchmarking/ claims it shows them in the console, and saves them to the file hl2hl2source.csv, Aha! wine steam.exe -login myuser mypass -applaunch 220 -novid -console -window -w 1024 -h 768 -dxlevel 90 -condebug +timedemoquit mydemo.dem will probably do what I want!
Re: [PATCH] ntdll: Show one-time ERR if NOFILE limit is reached.
Forgot to mention, this fixes http://bugs.winehq.org/show_bug.cgi?id=19861. Octavian
Re: [PATCH] ntdll: Show one-time ERR if NOFILE limit is reached.
On Sat, May 15, 2010 at 1:55 AM, Octavian Voicu octavian.vo...@gmail.com wrote: Forgot to mention, this fixes http://bugs.winehq.org/show_bug.cgi?id=19861. I also have another idea. There are many issues that depend on packagers to properly package wine, it's dependencies, and also configure various aspects that might influence its stability. There should be a page on the wiki with issues that package maintainers should be aware of. It should be structured something like General issues and Distribution-specific issues, with subcategories for each distribution, and a way to contact the maintainer. This is especially important with wine1.2 around the corner. Such a page could include, for example, a warning about this bug and tips for packagers on how to prevent it. Not sure about other distros, but in Ubuntu and other Debian-based ones the wine package could install a file /etc/security/limits.d/wine.conf to set the limit to a safe value. Or it could be in a separate package recommended by wine, like wine-nolimit-fix. Octavian