August 2007 - Wine-Devel - list.winehq.org

Re: [2/2] winex11: Don't use context-specific info when initializing

by Christoph Frick

On Fri, Aug 10, 2007 at 02:29:49AM -0700, Chris Robinson wrote: @@ -137,7 +132,7 @@ typedef struct wine_glpbuffer { } Wine_GLPBuffer; static Wine_GLContext *context_list; -static struct WineGLInfo WineGLInfo = { 0 }; +static struct WineGLInfo WineGLInfo; static int use_render_texture_emulation = 1; static int use_render_texture_ati = 0; static int swap_interval = 1; why is this changed? its a Good Thing to have global structs initialized so they fail earlier under abuse. -- cu

Aug. 10, 2007

Re: [MSI/tests] ConvertSidToStringSid not available on win98/NT4

by Paul Vriens

Jakob Eriksson wrote: > Kai Blin wrote: >> On Thursday 09 August 2007 18:01:47 Jakob Eriksson wrote: >> >> >>> My $.05 ; I would like to see the NT4 column all green some day, because >>> NT4 ended as a very stable and useful version of Windows. >>> If we can't even verify we support NT4 functionality with Wine, how far >>> have we really come? >>> >> That won't happen. E.g. I plan to eventually have Kerberos support in >> secur32.dll. NT4 doesn't do that. So whatever you do that's Win2k or beyond >> won't work on older Windows versions. That doesn't mean the test is bad. It >> also doesn't mean Wine can't run NT applications. >> >> I don't see how exactly having all test work on NT4 will make Wine better. >> > > Thats not exactly what I meant, I think. This Kerberos support comes > from extra functions > in secur32.dll right? The tests then will notice that there is no > Kerberos support and not > run Kerberos tests. Am I mistaken? > > Ahh.. :-) > > I did not mean a column with all green, but a column with no red. Quite > a difference. > Is that possible to achieve? > > > regards, > Jakob > > > > And that what I'm trying to get at as well. A test results page which is not red (and probably with a huge number of blue markings). Getting there is just like implementing the win32 api (chasing a running target). Tests change all the time. And people who implement new features use of course (?) winxp or higher to test. Which brings me to anther question. When are we defaulting to winxp? The test results look a bit better on xp than win2k. This doesn't mean of course we are doing better (haven't looked at the todo_wine's in the tests). Cheers, Paul.

Aug. 10, 2007

Re: [MSI/tests] ConvertSidToStringSid not available on win98/NT4

by Jakob Eriksson

Kai Blin wrote: > On Thursday 09 August 2007 18:01:47 Jakob Eriksson wrote: > > >> My $.05 ; I would like to see the NT4 column all green some day, because >> NT4 ended as a very stable and useful version of Windows. >> If we can't even verify we support NT4 functionality with Wine, how far >> have we really come? >> > > That won't happen. E.g. I plan to eventually have Kerberos support in > secur32.dll. NT4 doesn't do that. So whatever you do that's Win2k or beyond > won't work on older Windows versions. That doesn't mean the test is bad. It > also doesn't mean Wine can't run NT applications. > > I don't see how exactly having all test work on NT4 will make Wine better. > Thats not exactly what I meant, I think. This Kerberos support comes from extra functions in secur32.dll right? The tests then will notice that there is no Kerberos support and not run Kerberos tests. Am I mistaken? Ahh.. :-) I did not mean a column with all green, but a column with no red. Quite a difference. Is that possible to achieve? regards, Jakob

Aug. 9, 2007

Re: winex11: added missing sync_current_drawable calls

by Kai Blin

On Thursday 09 August 2007 15:59:59 plams wrote: > Here's a patch to fix a regression since 0.9.28 (in response to bug > 9029). OpenGL scissoring had apparently been wrapped in commit > 715d0e84097cb2dc07369c4e1cec8cf8a41e9035 but it seems that it made it > possible to make a glDisable(GL_SCISSOR_TEST) call that would leave GL > in some pseudo-state where the scissor test was STILL active (and vice > versa for glEnable) until e.g. a wglMakeCurrent or wglViewport call > was made. This patch should fix that (and thus resolve bug 9029). You need to provide your real name for the change log, otherwise your patch can't be accepted. Cheers, Kai -- Kai Blin WorldForge developer http://www.worldforge.org/ Wine developer http://wiki.winehq.org/KaiBlin Samba team member http://www.samba.org/samba/team/ -- Will code for cotton.

Aug. 9, 2007

Re: [MSI/tests] ConvertSidToStringSid not available on win98/NT4

by Kai Blin

On Thursday 09 August 2007 18:01:47 Jakob Eriksson wrote: > My $.05 ; I would like to see the NT4 column all green some day, because > NT4 ended as a very stable and useful version of Windows. > If we can't even verify we support NT4 functionality with Wine, how far > have we really come? That won't happen. E.g. I plan to eventually have Kerberos support in secur32.dll. NT4 doesn't do that. So whatever you do that's Win2k or beyond won't work on older Windows versions. That doesn't mean the test is bad. It also doesn't mean Wine can't run NT applications. I don't see how exactly having all test work on NT4 will make Wine better. Kai -- Kai Blin WorldForge developer http://www.worldforge.org/ Wine developer http://wiki.winehq.org/KaiBlin Samba team member http://www.samba.org/samba/team/ -- Will code for cotton.

Aug. 9, 2007

Re: ANNOUNCE: Wine/Mono Regression Test Suite

by Matt Jones

On 8/7/07, Dan Kegel <dank(a)kegel.com> wrote: > Matt wrote: > > ...I've found that running the mono regression test suite has > > given me a better idea of how to approach debugging mono/wine. > > ... > > http://mhjones.org/wine/wine-tests/wine-test-reader.tar.gz > > I downloaded this and had a look just now. > The instructions still refer to the deprecated 1.1 branch of > mono, and I hear to run on wine, one needs something newer > than the latest released version of mono. So I tried building > mono from the trunk. > > My dapper system was missing some things mono seems > to need, so I expanded on the script in the readme a bit, > and came up with > > #!/bin/sh > set -xe > MONOBUILDDIR=/data/dkegel/mono-cvs > > mcs --version || sudo apt-get install mono-devel > libtool --version || sudo apt-get install libtool > automake-1.9 --version || sudo apt-get install automake1.9 > automake --version | grep 1.9 || sudo update-alternatives --set > automake /usr/bin/automake-1.9 > You should also be able to call 'apt-get build-dep mono mono-devel' This should install everything you need > mkdir -p $MONOBUILDDIR > cd $MONOBUILDDIR > svn co svn://anonsvn.mono-project.com/source/trunk/mono > cd mono > ./autogen.sh --build i486-linux-gnu > make You should also call 'svn co svn://anonsvn.mono-project.com/source/trunk/mcs' immediately after checking out mono. > Can you update your README to show how to build some flavor > of mono-1.2 that runs under Wine? This should be up to date at some point later today. Thanks for the response, -matt

Aug. 9, 2007

Re: Wine Kernel Handle Support Module

by Michael Ost

>>> I know there is some discussion of what should be put in the kernel >>> module, so I am asking for advice on what to implement... ie. just >>> handles, everything wineserver does, etc. >> The first question is: >> Why do you want to do that and what specifically do >> you want to achieve? > > A small step towards less overhead in client <-> server calls? One _big_ vote for putting thread sync primitives --- events, semaphores, mutexes and critical sections --- in. We had to build a custom Wine to get acceptable realtime performance (thanks, Ulrich!). The client <-> server were causing high priority threads to block on low priority ones. I guess a test case would be something that shows how a high priority thread blocking on a sync primitive, while some low priority thread is doing some unrelated server heavy stuff, like lots of graphics updates. Cheers... mo

Aug. 9, 2007

Re: [MSI/tests] ConvertSidToStringSid not available on win98/NT4

by Jakob Eriksson

Paul Vriens wrote: > Hi, > > ConvertSidToStringSid (advapi32) is not available on win98/NT4 so the > msi tests (all of them) do not run on those platforms currently. > > I could use skip but that basically would mean no > MsiQueryProductState, MsiQueryFeatureState and MsiQueryComponentState > tests. > > Another thing is to copy over the ConvertSidToStringSid implementation > from advapi32 to the relevant tests. > > Any other ideas (or better yet, solutions)? My $.05 ; I would like to see the NT4 column all green some day, because NT4 ended as a very stable and useful version of Windows. If we can't even verify we support NT4 functionality with Wine, how far have we really come? (I know Wine is supposed to "run win32 applications", not "emulate a Windows version", but still...) regards, Jakob

Aug. 9, 2007

Re: [SNMPAPI] How to fix the tests for win98 and NT4?

by Jakob Eriksson

Paul Vriens wrote: > Hi, > > There are several functions in the util.c test that are not available > on win98 and NT4. That can be dealt with easily. > > Almost all tests with a NULL parameter crash on win98/NT4. I know that > a few weeks back we talked about ignoring win98 if it would mean > skipping a lot of tests, but now NT4 is involved as well. > > I could introduce a variable of course that's set when we are on NT4 > or lower (because of the missing functions) and don't run the 'NULL' > tests. > > Any ideas, comments? I had this idea rejected before. Something about the test should not test Windows version, only Windows behaviour, except when an app relies on it. (In which case it IS some kind of meta-behaviour one might say.) regards, Jakob

Aug. 9, 2007

[MSI/tests] ConvertSidToStringSid not available on win98/NT4

by Paul Vriens

Hi, ConvertSidToStringSid (advapi32) is not available on win98/NT4 so the msi tests (all of them) do not run on those platforms currently. I could use skip but that basically would mean no MsiQueryProductState, MsiQueryFeatureState and MsiQueryComponentState tests. Another thing is to copy over the ConvertSidToStringSid implementation from advapi32 to the relevant tests. Any other ideas (or better yet, solutions)? Cheers, Paul.

Aug. 9, 2007