June 2008 - Wine-devel - list.winehq.org

Re: kernel32: initialize hole in struct in WaitNamedPipeW to avoid Valgrind warnings, take 2 (resend)
by Alexandre Julliard 23 Jun '08

23 Jun '08

"Dan Kegel" <dank(a)kegel.com> writes: > Back in February, I tried doing this with a real memset, > but Alexandre demurred, probably objecting to the slight overhead. > So here's a version with no overhead unless you > have the Valgrind development headers installed, > and is done with Valgrind client notations. This isn't the right place for this, and pretty much all ioctls have the same potential problem. You should add a valgrind suppress for the whole category of such problems. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [1/2] wined3d: Report post pixelshader blending on WINED3DFMT_A16B16G16R16F
by Roderick Colenbrander 23 Jun '08

23 Jun '08

Hi, I don't think we should implement it this way. For instance on GeforceFX cards blending is performed in software. Not too long ago (before 1.0) I also sent a patch for this e.g. by doing an extension check. I think I checked for ARB_color_buffer_float in combination with half pixel support. Roderick -- Ist Ihr Browser Vista-kompatibel? Jetzt die neuesten Browser-Versionen downloaden: http://www.gmx.net/de/go/browser

2 2

RPC questions
by Kai Blin 23 Jun '08

23 Jun '08

Hi folks, hey Rob, I've been wanting to get into Wine RPC development for ages, and now I finally got myself to start reading through some of the code in dlls/rpcrt4. Now, after going over rpc_bindings.c, tests/rpc.c and tests/server.c, I've got a couple of questions about that code. I'd like to add a test for RpcBindingSetAuthInfo. It seems like server.c is the place to put it, right? If I want to use RPC_C_AUTHN_WINNT, I need server side authentication for this test. Has anyone ever tried that before? I'm assuming I might need to run Wine as root in order to be able to connect to the winbindd privileged pipe in order to check the user, need to try that in "plain" ntlm_auth first. Rob, you said you set up RPC so requests for Negotiate would be thumped to NTLM. I can't see that in RpcBindingSetAuthInfo. Is that patch in the tree, or is it too hacky? 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.

2 1

today's valgrind results online
by Dan Kegel 23 Jun '08

23 Jun '08

Online at http://kegel.com/wine/valgrind/20071002/ (I now include more info about memory leaks, but I still only report anything if there was an invalid memory reference.) The following tests were fixed in git yesterday, and no longer have any valgrind warnings: comctl32/rebar comctl32/status crypt32/chain crypt32/crl crypt32/msg cryptnet/cryptnet user32/msg wintrust/softpub Juan thinks nearly all of the remaining crypt32 errors can be suppressed (some are intentional). Here are the directories that still have warnings: advapi32 advpack cabinet comctl32 crypt32 d3d8 d3d9 d3dx8 dplayx dsound gdi32 gdiplus hlink kernel32 mlang msacm32 mscms mshtml msi msvcrt msxml3 netapi32 ntdll ole32 oleaut32 quartz riched20 riched32 rpcrt4 rsaenh setupapi shdocvw shell32 urlmon user32 usp10 wininet winmm wintrust - Dan

3 4

re: richedit: WM_SETTEXT must immediately autodetect URLs, without waiting for a WM_CHAR.
by Dan Kegel 23 Jun '08

23 Jun '08

Hi Alex! I just noticed a valgrind warning in richedit that might have something to do with your new test code in http://source.winehq.org/git/wine.git/?a=commit;h=38d7ba6eff6757b83b54b0485… The warning is: Conditional jump or move depends on uninitialised value(s) at test_EM_AUTOURLDETECT (editor.c:1559) by func_editor (editor.c:4203) by run_test (test.h:449) by main (test.h:498) Uninitialised value was created by a client request at mark_block_uninitialized (heap.c:164) by RtlAllocateHeap (heap.c:1239) by heap_alloc (editor.h:28) by RichEditWndProc_common (editor.c:2557) by RichEditWndProcA (editor.c:3353) by ??? (library.h:163) by call_window_proc (winproc.c:457) by WINPROC_call_window (winproc.c:2207) by call_window_proc (message.c:1610) by send_message (message.c:2434) by SendMessageA (message.c:2579) by test_EM_AUTOURLDETECT (editor.c:1545) by func_editor (editor.c:4203) by run_test (test.h:449) by main (test.h:498) Could you have a look at it? Thanks, Dan

1 0

Re: msxml3/tests: VT_EMPTY is not a string (resend 2)
by Paul Vriens 23 Jun '08

23 Jun '08

Dan Kegel wrote: > This time using NULL instead of (void *) 0. > Hi Dan, - ok(lstrlenW(V_BSTR(&vValue)) == 0, "Expected empty value\n"); + ok(V_BSTR(&vValue) == (void *)0, "Expected null value\n"); Still using (void *)0 though. -- Cheers, Paul.

1 0

re: programs: add calc.exe
by Dan Kegel 23 Jun '08

23 Jun '08

> I see he has taken the page down. Maybe drop the guy an email and see > if he still has the sources laying around. A little google searching found it's still there, just moved: http://jebriggs.com/winecalc/ Source is all there, about 5KLOC. - Dan

1 0

Re: programs: add calc.exe
by Steven Edwards 22 Jun '08

22 Jun '08

On Sun, Jun 22, 2008 at 10:32 AM, Louis. Lenders <xerox_xerox2000(a)yahoo.co.uk> wrote: > Hi, i noticed this while running Log Book Pro (http://logbookpro.com/) > That program has a menuentry to start Windows calculator, but it does > nothing, as wine does not implement this. With the attached patch at least a > calculator is started, instead of doing nothing. There was a Calc knock off sent to us a while back, you might try merging it in instead http://www.winehq.org/pipermail/wine-patches/2004-January/009481.html -- 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

1 1

Website down
by Austin English 22 Jun '08

22 Jun '08

Did we get slashdotted/dugg to death? -Austin

6 6

Re: msi [6/6]: Rewrite the PublishProduct test to only check the registry changes
by Paul Vriens 22 Jun '08

22 Jun '08

James Hawkins wrote: > Hi, > > Changelog: > * Rewrite the PublishProduct test to only check the registry changes. > > dlls/msi/tests/install.c | 148 +++++++++++++++++++++++++++------------------- > 1 files changed, 87 insertions(+), 61 deletions(-) > > > > ------------------------------------------------------------------------ > > Hi James, I've noticed some (new) failure for the msi/install test: + CHECK_DEL_REG_STR(hkey, "Clients", ":"); + } This one is a REG_MULTI_SZ on my W2K3 box (and from the looks of the test results probably the same on XP/W2K3). The macro CHECK_DEL_REG_STR or actually the function check_reg_str checks only for REG_SZ and REG_EXPAND_SZ. I'm not sure what the fix should be: - replace the check for REG_EXPAND_SZ by REG_MULTI_SZ - add the REG_MULTI_SZ as well to the check -- Cheers, Paul.

2 1