May 2010 - Wine-devel - list.winehq.org

performance bot results - this time with screensaver off
by Dan Kegel 20 May '10

20 May '10

Every benchmark did at least a little better, and 3dmark2001 did a lot better, today: http://kegel.com/wine/yagmarkdata/E8400-GT_220-Ubuntu_10.04_LTS-e8400-wine-… Why? Well, there are oodles of processes running in Gnome by default, plus the workrave app I use for my RSI. I had killed some of them on the benchmark machine earlier, but forgot to kill them again after rebooting; just noticed that last night. Today's results are with three more processes killed: workrave, gnome-screensaver, and bluetooth-applet. That might help explain the results a bit. I did 'sudo apt-get install xubuntu-desktop' just now, rebooted, edited ~/.pulse/client.conf to disable pulseaudio autospawn, and killed pulseaudio xscreensaver update-notifier /usr/share/system-config-printer/applet.py /usr/lib/gvfs//gvfs-fuse-daemon bluetooth-applet There are now 'only' 148 or so processes in my idle system, only 29 of which are the interactive user; it seems fairly stripped down. I also ran nvidia-settings again to pick 'maximum performance' rather than adaptive, and told xfce to never blank the monitor. Let's see if all that makes a difference. (I need to make a big checklist of Things To Do After Rebooting.) - Dan

1 0

The WineHQ About page needs a picture
by Scott Ritchie 19 May '10

19 May '10

I was doing some housecleaning work looking at the WineHQ website, and I realized we still have an awful lot of flat, wide text on the About page. This is the perfect place to collapse it into a column and fill the right side of the screen with an image. But...what image? Thanks, Scott Ritchie

3 2

re: wininet: Combine certificate errors in HTTP_OpenConnection
by Dan Kegel 19 May '10

19 May '10

>> Test can't be easily added to wine because we don't have pages to test >> certificate errors on. > >That should be addressed, no? E.g. with a page hosted at winehq.org >with a self-signed certificate? Or look at how Chromium does its ssl unit tests, by launching a little server written in python: http://src.chromium.org/cgi-bin/gitweb.cgi?p=chromium.git;a=blob;f=net/sock… http://src.chromium.org/cgi-bin/gitweb.cgi?p=chromium.git;a=blob;f=net/sock… - Dan

3 3

Re: [PATCH 4/4] wininet: Fix variables values after redirection in HTTP_HttpSendRequestW
by Alexandre Julliard 19 May '10

19 May '10

Piotr Caban <piotr(a)codeweavers.com> writes: > --- > dlls/wininet/http.c | 4 ++++ > 1 files changed, 4 insertions(+), 0 deletions(-) It doesn't work here: ../../../tools/runtest -q -P wine -M wininet.dll -T ../../.. -p wininet_test.exe.so http.c && touch http.ok http.c:405: Test failed: expected status 10 (INTERNET_STATUS_RESOLVING_NAME) 1 times, received 2 times http.c:406: Test failed: expected status 11 (INTERNET_STATUS_NAME_RESOLVED) 1 times, received 2 times make: *** [http.ok] Error 2 -- Alexandre Julliard julliard(a)winehq.org

1 0

Heaven 2 benchmark in good shape...
by Dan Kegel 19 May '10

19 May '10

Yesterday's performance bot results, http://kegel.com/wine/yagmarkdata/E8400-GT_220-Ubuntu_10.04_LTS-e8400-wine-… show that the Heaven 2 d3d9 benchmark had a couple 10% improvements in the past week (up from 10 to 11 and then 12 fps). If we're going to see improvements in a benchmark, that's the best one, since it probably matches modern games more closely than the others. 3dmark2001 had a very good day (http://kegel.com/wine/yagmarkdata/wine-1.1.44-19-vs-wine-1.1.44-72.txt) but returned to its poor scores thereafter. No idea why the one-day blip. (There was also an upward blip in heaven2_gl that day, but it was quite small.)

1 0

Re: d3dx9: Add some version-specific checks in the shader assembler.
by Stefan Dösinger 19 May '10

19 May '10

Am 18.05.2010 um 14:55 schrieb Matteo Bruni: > + if(!check_reg_type(src, vs_3_reg_allowed)) { > + asmparser_message(This, "Line %u: Source register %s not supported in VS 3.0\n", > + This->line_no, > + debug_print_srcreg(src, ST_VERTEX)); > + set_parse_status(This, PARSE_ERR); > + } I realize the patch has been committed already, but I'm wondering: Does the native assembler check the register numbers too? Or just the types? You have to be careful with error checking in the assembler, because the native assembler checks for very few error conditions. More complex checks like uninitialized registers or unwritten output registers are checked by the shader validator, an undocumented interface in d3d9.dll(which is called by vsa.exe / psa.exe) If my memory is correct, all the other error conditions you check for are checked by the native *assembler*, but I'm not sure about the register numbers. I think more unit tests would be helpful.

2 1

Re: winemp3.acm: Fixes the Mac-specific code (try 2)
by Aric Stewart 19 May '10

19 May '10

Hi, I was excited to try out this patch but now it is giving me a crash. (OS/X 10.5.8) Backtrace: =>0 0xffff12a9 (0x0033d81c) 1 0xffff086e (0x0033d838) 2 0x936f5a47 _ZN14CodecConverter14AppendNewInputEmPcmRmP28AudioStreamPacketDescription+0xb6() in audiotoolbox (0x0033d888) 3 0x936f564b _ZN14CodecConverter13CallInputProcERm+0x1ca() in audiotoolbox (0x0033d8f8) 4 0x936f513d _ZN14CodecConverter17DecoderFillBufferERmR15AudioBufferListP28AudioStreamPacketDescription+0x47c() in audiotoolbox (0x0033d9c8) 5 0x936e9b61 _ZN22BufferedAudioConverter13GetInputBytesEmRmRPK12CABufferList+0xa0() in audiotoolbox (0x0033d9f8) 6 0x936e9a1d _ZN12CBRConverter12RenderOutputEP12CABufferListmRmP28AudioStreamPacketDescription+0x7c() in audiotoolbox (0x0033da38) 7 0x936e978c _ZN22BufferedAudioConverter10FillBufferERmR15AudioBufferListP28AudioStreamPacketDescription+0xeb() in audiotoolbox (0x0033da98) 8 0x936e9b61 _ZN22BufferedAudioConverter13GetInputBytesEmRmRPK12CABufferList+0xa0() in audiotoolbox (0x0033dac8) 9 0x936e9a1d _ZN12CBRConverter12RenderOutputEP12CABufferListmRmP28AudioStreamPacketDescription+0x7c() in audiotoolbox (0x0033db08) 10 0x936e978c _ZN22BufferedAudioConverter10FillBufferERmR15AudioBufferListP28AudioStreamPacketDescription+0xeb() in audiotoolbox (0x0033db68) 11 0x936e9914 _ZN19AudioConverterChain12RenderOutputEP12CABufferListmRmP28AudioStreamPacketDescription+0x63() in audiotoolbox (0x0033dba8) 12 0x936e978c _ZN22BufferedAudioConverter10FillBufferERmR15AudioBufferListP28AudioStreamPacketDescription+0xeb() in audiotoolbox (0x0033dc08) 13 0x93700b23 AudioConverterFillComplexBuffer+0x112() in audiotoolbox (0x0033dc58) 14 0x57b4f560 _mp3_leopard_horse+0x2bf() in winemp3 (0x0033dd08) 15 0x57b4f9a5 _MPEG3_DriverProc+0xf4() in winemp3 (0x0033ddd8) Here is the mpeg log snipit from just before the crash: trace:mpeg3:MPEG3_DriverProc (00000000 0x27af350 0003 00000000 0033de0c); trace:mpeg3:MPEG3_DriverProc (00000001 0x27af350 601b 0033dea8 00000000); trace:mpeg3:MPEG3_DriverProc (00000000 0x1114418 0003 00000000 0033dd9c); trace:mpeg3:MPEG3_DriverProc (00000001 0x1114418 604c 008f757c 00000000); trace:mpeg3:MPEG3_DriverProc (00000001 0x1114418 604e 008f757c 0033deb0); trace:mpeg3:MPEG3_DriverProc (00000001 0x1114418 6051 008f757c 0033df44); trace:mpeg3:MPEG3_DriverProc (00000001 0x1114418 604f 008f757c 0033df44); trace:mpeg3:mp3_leopard_horse ndst 147456 0x3127860 <- 19901 0x3122a94 trace:mpeg3:mp3_leopard_horse First 16 bytes to input: "\x00\x1f\x84\n\x92_\x81h\x00\x00\x00\x00\x00\x00\x00\x00" I am also getting quite a number of build warning that i dont know if should be concerning. Anything more I can do to test for you? -aric Kristofer Henriksson wrote: > Second submission, which improves the previous patch by properly > syncing to the initial MP3 frame header. > > The current winemp3.acm does not work if the Mac-specific code is > used. This rewrites the relevant portion to fix, for example, > background music on things like Oblivion. > > > ------------------------------------------------------------------------ > >

1 0

Re: Work on WineQuartz for MacOSX
by James Mckenzie 19 May '10

19 May '10

Roderick Colenbrander wrote: >On Sun, May 16, 2010 at 6:02 PM, C.W. Betts <computers57(a)hotmail.com> wrote: >> On May 16, 2010, at 9:31 AM, James Mckenzie wrote: >>> All: >>> >>> There was or is a quite lengthy thread on this project. >>> >>> What is the status of work? Any progress on Emmanuel's work from 2008? >>> >>> Thank you. >>> >>> James McKenzie >>> >> If memory serves, the main reason why there hasn't been any work done is because of the >> dependency of Objective-C, which the project administrator has said "no" to. >> >The fact that Cocoa is objective C is one of the issues (Charles Davis >made some C wrapper I think). The most problematic thing is that it >really need is a DIB engine. The Cocoa graphics APIs are similar to >Cairo / XRender / Direct2D and other modern APIs. These APIs are great >for vector drawing, alpha blending and other fancy operations but they >are not low-level 2D APIs. For instance they lack most classic >(bitwise) 2D ROPs. The classic rendering stuff (Windows 7 does it this >way as well) has to be performed in software by a DIB engine. > >Roderick > Charles Davis did not 'make a c wrapper', Apple did. However you are more correct in that we need a fully functional DIB engine in order to fully move the Apple interface from X11 to Aqua. I was inquiring if both of these actions have occurred. Max was working diligently on a DIB engine until AJ basically shot his work full of holes and he left. Again, what is the status of updating/upgrading Emmanuel's work? Has it been left for dead or is someone working in the background on it? I would like to pick it up and start looking at it again and seeing if the Obj-C code could be converted to plain 'c' or if we have to bring up a sidebar version of Wine that will be Aqua complaint. Having been through this with the OpenOffice.org and NeoOffice.org projects, this is a political area I really don't want to visit, but if it has to be that way, then it will head off in that direction. James McKenzie

4 4

Re: [PATCH 2/4] wininet: Added INTERNET_OPTION_ERROR_MASK flag handling to InternetSetOptionW (try2)
by Austin English 19 May '10

19 May '10

On Tue, May 18, 2010 at 8:28 PM, Piotr Caban <piotr(a)codeweavers.com> wrote: > Try2: > - Commented out test that fails on Win2000 (there were no new failures on > Win98). Any reason you didn't use broken() instead? -- -Austin

1 0

Re: ntdll: add stub for NtRaiseHardError (try 2)
by James McKenzie 19 May '10

19 May '10

Austin English wrote: > On Tue, May 18, 2010 at 1:17 PM, Austin English <austinenglish(a)gmail.com> wrote: > >> This time returning STATUS_NOT_IMPLEMENTED instead of E_NOTIMPL, to >> match the rest of ntdll. Thanks to Juan for pointing that out. >> > > Today's not my day... > > Wasn't mine either. Spent the entire day at 'work' chasing an error. Finally figured it out 1/2 hour before I left. Now to fix 140,000 or so database entries. James McKenzie

1 0