September 2008 - Wine-Devel

Re: imm32: Improve ImmGetCompositionString.
by Paul Vriens Sept. 8, 2008

Sept. 8, 2008

Aric Stewart wrote: > > It should return number of copied bytes. But it always returns required > buffer size to receive all information. > (originally by Kusanagi Kouichi (slash(a)ma.neweb.ne.jp) ) > --- > dlls/imm32/imm.c | 207 > ++++++++++++++++++++++++++++++++++-------------------- > 1 files changed, 130 insertions(+), 77 deletions(-) > > > > ------------------------------------------------------------------------ > > Shouldn't patches like these be accompanied by tests? It changes the behavior after all. -- Cheers, Paul.

2 1

Re: dsound: fix a few failing tests and a crash on 2008
by Paul Vriens Sept. 8, 2008

Sept. 8, 2008

Jeff Zaroyko wrote: > > ------------------------------------------------------------------------ > > I found more or less the same thing for Vista a few days ago. Doing a IDirectSoundBuffer_Release sets the reference counter to zero but doesn't clear secondary. Is this something worthwhile to add a comment for (or a test)? It does seem important as it's a change of behavior starting with Vista? -- Cheers, Paul.

2 1

Re: winemenubuilder: eliminate wineshelllink
by Francois Gouget Sept. 8, 2008

Sept. 8, 2008

On Thu, 4 Sep 2008, Damjan Jovanovic wrote: > Changelog: > * moved the functionality of wineshelllink into winemenubuilder, and > eliminated wineshelllink What is the rational for eliminating wineshelllink? It seems to me that by eliminating it we are losing a lot of flexibility for handling special setups (and there are a lot of these in menuing systems). -- Francois Gouget <fgouget(a)free.fr> http://fgouget.free.fr/ Research is the transformation of money to knowledge. Innovation is the transformation of knowledge to money. -- Dr. Hans Meixner

4 4

Re: crypt32: fixed the testGetCertChain tests on Vista.
by James Hawkins Sept. 8, 2008

Sept. 8, 2008

On Sun, Sep 7, 2008 at 5:59 AM, Reece Dunn <msclrhd(a)googlemail.com> wrote: > This fixes these crypt32:chain testGetCertChain tests on Vista and Windows 2008. > - ok(!ret && GetLastError() == ERROR_INVALID_DATA, - "Expected ERROR_INVALID_DATA, got %d\n", GetLastError()); + ok(!ret, "Expected CertGetCertificateChain to fail\n"); + ok(GetLastError() == ERROR_INVALID_DATA || + GetLastError() == CRYPT_E_ASN1_BADTAG /* Vista/win2k8 */, I'm not NACK'ing this patch, but for future reference, please put comments *after* the comma or logical OR marker, and each system condition should be on its own line. I've seen this in several patches recently. While it's syntactically OK, it looks wrong and there's no need for everyone to have to play Find The Comma. Yes there are other places in the code where this happens, but that doesn't make it correct or preferable. For example, the following is significantly easier to read: ok(result == ERROR_BUFFER_TOO_SMALL || result == ERROR_INVALID_USER_BUFFER || /* win98 */ result == ERROR_INVALID_DATA, /* Vista */ "Expected ERROR_BUFFER_TOO_SMALL, got %08d\n", result); than ok(result == ERROR_BUFFER_TOO_SMALL || result == ERROR_INVALID_USER_BUFFER /* win98 */ || result == ERROR_INVALID_DATA /* Vista */, "Expected ERROR_BUFFER_TOO_SMALL, got %08d\n", result); -- James Hawkins

2 3

Re: user32: Resend Check for NULL pData in DdeClientTransAction expectingto be passed a handle
by Dmitry Timoshkov Sept. 8, 2008

Sept. 8, 2008

"Jeff Latimer" <lats(a)yless4u.com.au> wrote: > diff --git a/dlls/user32/dde_client.c b/dlls/user32/dde_client.c > index 99e2d1c..cbbf99a 100644 > --- a/dlls/user32/dde_client.c > +++ b/dlls/user32/dde_client.c > @@ -632,6 +632,15 @@ static WDML_XACT* WDML_ClientQueueExecute(WDML_CONV* pConv, LPVOID pData, DWORD > > TRACE("XTYP_EXECUTE transaction\n"); > > + if (pData == NULL) > + { > + if (cbData == (DWORD)-1) > + pConv->instance->lastError = DMLERR_INVALIDPARAMETER; > + else > + pConv->instance->lastError = DMLERR_MEMORY_ERROR; > + return NULL; > + } Alexandre suggested to use the existing 'if (cbData == (DWORD)-1)' block, not introduce a new one. > @@ -1154,13 +1163,10 @@ HDDEDATA WINAPI DdeClientTransaction(LPBYTE pData, DWORD cbData, HCONV hConv, HS > switch (wType) > { > case XTYP_EXECUTE: > - /* Windows simply ignores hszItem and wFmt in this case */ > - if (pData == NULL) > - { > - pConv->instance->lastError = DMLERR_INVALIDPARAMETER; > - return 0; > - } > + /* Windows simply ignores hszItem and wFmt in this case */ > pXAct = WDML_ClientQueueExecute(pConv, pData, cbData); > + if (pXAct == NULL) > + return 0; > break; The formatting of the above block is strange. Set the tab size in your editor to 8 and you will see the mess. -- Dmitry.

2 2

Fixing the conformance tests
by Paul Vriens Sept. 8, 2008

Sept. 8, 2008

Hi, As we are slowly progressing towards a more green test.winehq.org there are still some conformance tests that fail either on all platforms or with a specific configuration (like IE7). This email is merely to list those and see if the SME's can have a look in these tests. (I didn't put to much effort in stuff below XP and left out user32) Tests with timeouts =================== - dplayx:dplayx fails with a timeout on all windows servers - msi:install fails with a timeout on a lot of windows servers Tests that fail with IE7 installed ================================== - hlink:hlink has a consistent 8 failures on servers with IE7 - mshtml:htmldoc has 8 failures on servers with IE7 (and more, but also on others) - shdocvw:webbrowser has 3 failures on servers with IE7. Tests that fail starting with a specific OS (or level) ====================================================== - msi:package has 16 failures on XP-SP3+ and 16 extra on W2K3-SP2+ - ole32:storage32 has 24 failures on XP-SP3+ and 24 extra on W2K3-SP2+ (both the above have to do with access rights on files) Tests that fail on Windows and succeed on Wine ============================================== - ws2_32:sock has 1 consistent failure on all Windows boxes but not on Wine Tests with failures on all Windows boxes ======================================== - riched20:editor has failures on all boxes - urlmon:protocol either has timeouts or failures on all boxes - urlmon:url fails with 6 consistent failures on all boxes The above is a subset of all the failures and are not the easy ones. There are still loads of tests that fail on only 1 OS or even on 1 server. -- Cheers, Paul.

4 4

Re: itircl: add a stub for DllUnregisterServer
by Austin English Sept. 7, 2008

Sept. 7, 2008

Please ignore this patch. It's not ready, and not sure why it was sent yet... -Austin On Sun, Sep 7, 2008 at 2:08 PM, Austin English <austinenglish(a)gmail.com> wrote: > >

1 0

Re: [PATCH] buffer overflow checking for string functions
by Francois Gouget Sept. 7, 2008

Sept. 7, 2008

On Thu, 4 Sep 2008, Marcus Meissner wrote: [...] > +#if defined(__GNUC__) && (__GNUC__ < 4) > +# define __builtin_object_size(x,y) -1 > +#endif Shouldn't this be: #if !defined(__GNUC__) || (__GNUC__ < 4) To preserve compatibility with non-GNU compilers. > +static inline INT > +WINAPI MultiByteToWideChar_ichk( > + UINT cp,DWORD flags, > + LPCSTR src,INT srclen,INT srcbuflen, > + LPWSTR dst,INT dstlen,INT dstbuflen > +) { > + if ((srclen != -1) && (srcbuflen != -1) && (srcbuflen < srclen)) > + RaiseException(0xC0000409,0x01,0,NULL); It would be nice to issue a trace clarifying the cause of the exception before raising it. Maybe something like this: TRACE("used %d as the size when >= %d is needed for the builtin object\n", ...); Sending it to the seh channel would be best. [...] > +#define MultiByteToWideChar(cp,flags,src,srclen,dst,dstlen) \ > + MultiByteToWideChar_ichk(cp,flags, \ > + src,srclen,__builtin_object_size(src,0), \ > + dst,dstlen,__builtin_object_size(dst,0) \ > + ) I have a few other concerns here: * Why do we need a macro here? I thought it was so that __builtin_object_size() could do its work, but the strcpy() functions above in the patch have no associated macro and they call __builtin_object_size() on their parameter which would seem to be poinless. So I must be misunderstanding something here. * It's customary to add some extra parenthesing in such macros: (src),(srclen),__builtin_object_size((src),0), * The macro itself can cause trouble in some cases. In Winelib code it can cause naming collisions with user functions (especially with class methods, though it may not be very likely here), which would especially be an issue with strcpy*(). Maybe it should be protected by something like a WINE_NO_OVERFLOW_CHECK macro? * The macro can also cause trouble in case its parameters have side-effects, like x++ or similar (though the __builtin_object_size() mentions returning -1 in case or side-effects). This could impact Wine too. -- Francois Gouget <fgouget(a)free.fr> http://fgouget.free.fr/ Computers are like airconditioners They stop working properly if you open WINDOWS

3 2

Changing configure.ac script
by Nikolay Sivov Sept. 7, 2008

Sept. 7, 2008

Is there any way not to include a regenerated 'configure' script into a patch but be able to 'git fetch' it when it's changed remotely in repo? As I see there is no need to patch a 'configure' file in repository since it's always regenerated when 'configure.ac' changed. Thanks in advance.

2 4

1st resend: Can we import MSConfig from ReactOS? (was: autorun perhaps dangerous)
by Jason Spiro Sept. 7, 2008

Sept. 7, 2008

(Hi Alexandre. I sent the below email 2 weeks ago and didn't get a response yet. I know you are busy, but please do try to find the time to reply, even if it's to say "I'm not sure". Cheers, --Jason) cc: wine-devel(a)winehq.org Hi Alexandre, In January 2008, Steven Edwards wondered[1] if you would let Wine import "msconfig" (the System Configuration Utility) from ROS. Would you allow it? Thanks, --Jason Spiro (not a Wine developer) ^ [1]. http://news.gmane.org/find-root.php?message_id=%3c19e31a080801122302u52d513… -- Jason Spiro: software/web developer, trainer, IT consultant. I support Linux, UNIX, Windows, and more. Contact me to discuss your needs and get a free estimate. +1 (613) 668-6096 / info(a)jspiro.com

4 5