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

RFC: Implement XML loading in msxml3 directly from IStream
by Michael Karcher 22 Oct '08

22 Oct '08

Hello developers, the current implementation of IPersistStream::Load for XMLDOMDocument takes a copy of the given stream into a hglobal stream, and uses a helper function to parse that, namely parse_xml that calls xmlReadMemory if available (since 2.6.0) and uses the old xmlParseMemory call otherwise. If libxml2 is new enough, it is possible to parse directly from the stream using xmlReadIO to avoid copying the whole stream contents to memory. The function xmlReadIO was introduced in libxml 2.6.0, five years ago. If I want to use that function, do I have to create an alternative code path for such ancient libxml implemementations, or is it OK to require libxml 2.6.0 as minimum for libxml (even Debian Sarge, i.e. oldstable has it)? Another question arose on looking at the current implementation. If Wine is not going the direct-stream-to-libxml2-way, but wants to take a copy, currently the copying is performed manually, although there is a function CopyTo in IStream that takes care of creating the copy. I attached two diff files, - load-via-readio.diff shows the way that calls stream callbacks from libxml2 - load-use-copyto.diff uses the CopyTo function, but still copies. Thanks in advance for any feedback, Michael Karcher

1 1

Re: advapi32: Fall back to /dev/random if /dev/urandom can't be opened, and add warnings if opening either fails (try 2)
by Alexandre Julliard 22 Oct '08

22 Oct '08

"Juan Lang" <juan.lang(a)gmail.com> writes: > Changed a WARN to a FIXME. Any other suggestions? I don't see the point. Is there really a platform that has /dev/random but not /dev/urandom, and a /dev/random that is guaranteed not to block? -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: shell32: try to guess the working directory in the run dialog
by Juan Lang 22 Oct '08

22 Oct '08

Hi Vincent, + while (*src) + { + *dest = *src; + if (*src == '\\') + { + DWORD attrs = GetFileAttributesW(result); It sure looks like you're calling GetFileAttributesW with a non-NULL-terminated string. I doubt this will work as you intend. --Juan

3 5

Re: configure: Require at least version 1.6 of gnutls.
by James Mckenzie 22 Oct '08

22 Oct '08

Henri Verbeet <hverbeet(a)gmail.com> at Oct 21, 2008 10:04 AM wrote about configure: Require at least version 1.6 of gnutls. > > Why are we looking to use gnutls 1.6? Is there a feature that is needed for Wine? If so, I will have to check Fink before building Darwine again. Thank you. James McKenzie

2 1

Re: paint: Add new program
by Austin English 21 Oct '08

21 Oct '08

On Mon, Oct 20, 2008 at 8:27 PM, Hirofumi Katayama <katayama.hirofumi.mz(a)gmail.com> wrote: > See attachment. > > > > Howdy, Thanks for the patch, a few comments: >From b5fc5d2fc2c74507aa2dcf2daa4fe404e9b1ba06 Mon Sep 17 00:00:00 2001 From: Katayama Hirofumi MZ <katayama.hirofumi.mz(a)gmail.com> Date: Tue, 21 Oct 2008 03:41:08 +0900 Subject: [PATCH] paint: Add new program --- .gitignore | 1 + configure | 11 + configure.ac | 1 + Don't include changes to configure, it is built automatically from configure.ac. Make any needed changes there. Your patch doesn't apply cleanly: bash-3.2$ git apply ../paint.txt ../paint.txt:425: trailing whitespace. ../paint.txt:428: trailing whitespace. ../paint.txt:438: trailing whitespace. ../paint.txt:449: trailing whitespace. ../paint.txt:765: trailing whitespace. warning: configure has type 100755, expected 100644 warning: tools/make_makefiles has type 100755, expected 100644 warning: squelched 71 whitespace errors warning: 76 lines add whitespace errors. Why are you changing the permissions on those files? -- -Austin

1 0

Re: configure: Require at least version 1.6 of gnutls.
by Alexandre Julliard 21 Oct '08

21 Oct '08

"Henri Verbeet" <hverbeet(a)gmail.com> writes: > @@ -916,17 +916,20 @@ WINE_NOTICE_WITH(hal,[test "x$ac_cv_lib_soname_hal" = "x"], > dnl **** Check for libgnutls **** > if test "x$with_gnutls" != "xno" -a "$PKG_CONFIG" != "false" > then > - ac_save_CPPFLAGS="$CPPFLAGS" > - ac_gnutls_libs="`$PKG_CONFIG --libs gnutls 2>/dev/null`" > - ac_gnutls_cflags="`$PKG_CONFIG --cflags gnutls 2>/dev/null`" > - CPPFLAGS="$CPPFLAGS $ac_gnutls_cflags" > - AC_CHECK_HEADER(gnutls/gnutls.h, > - [WINE_CHECK_SONAME(gnutls,gnutls_global_init, > - [AC_SUBST(GNUTLSINCL,"$ac_gnutls_cflags")],,[$ac_gnutls_libs])]) > - CPPFLAGS="$ac_save_CPPFLAGS" > + if $PKG_CONFIG --atleast-version=1.6 gnutls You should check for the features we depend on, not the version number. -- Alexandre Julliard julliard(a)winehq.org

1 0

Patchwatcher status
by Dan Kegel 21 Oct '08

21 Oct '08

The blacklist has grown a bit, adding d3d8:visual.c winhttp:Timeout...Killing.child wldap32:parse.c wldap32/tests/__test__ Rob has been busy helping get the distributed patchwatcher scripts working. He has a msvc9 build slave coming along nicely(!). I've ordered a second computer to use as a second build slave, and am considering how to get the latency of valgrind runs low enough to be useful. (I think I've going to add a filter to the alarm.c I pass as WINETEST_WRAPPER, and use that to select which part of the test suite each build slave should run. That way I can have five build slaves each doing a fifth of the test suite in parallel.) It's not quite trivial to get this all going, so no idea when it'll be online. - Dan

1 0

Re: winmm: MMIO_ALLOCBUF is ignored if mmioOpen gets a valid buffer.
by Paul Vriens 21 Oct '08

21 Oct '08

Dmitry Timoshkov wrote: > Existing tests show that MMIO_ALLOCBUF flag is dropped if a valid buffer > is being passed to mmioOpen, that means the buffer is valid in that case. > > This patch fixes a regression reported in the bug 15694. > --- > dlls/winmm/mmio.c | 5 +++-- > 1 files changed, 3 insertions(+), 2 deletions(-) > > diff --git a/dlls/winmm/mmio.c b/dlls/winmm/mmio.c > index 60b6d0f..a750bf2 100644 > --- a/dlls/winmm/mmio.c > +++ b/dlls/winmm/mmio.c > @@ -667,10 +667,11 @@ HMMIO MMIO_Open(LPSTR szFileName, MMIOINFO* refmminfo, DWORD dwOpenFlags, > refmminfo->wErrorRet = MMIO_SetBuffer(wm, refmminfo->pchBuffer, refmminfo->cchBuffer, 0); > if (refmminfo->wErrorRet != MMSYSERR_NOERROR) > goto error1; > - if (wm->info.fccIOProc == FOURCC_MEM) > - wm->bBufferLoaded = TRUE; > } > > + if (wm->info.fccIOProc == FOURCC_MEM && !(wm->info.dwFlags & MMIO_ALLOCBUF)) > + wm->bBufferLoaded = TRUE; > + > /* see mmioDosIOProc for that one */ > wm->info.adwInfo[0] = refmminfo->adwInfo[0]; > Hi, Is there a way the existing tests can be improved? Apparently the existing tests show the behaviour but didn't prevent the regression. -- Cheers, Paul.

2 1

Re: wldap32/tests: modify test that was failing on every windows system
by Hans Leidekker 21 Oct '08

21 Oct '08

On Tuesday 21 October 2008 14:29:45 Aric Stewart wrote: > - ok( !ret, "ldap_search_ext_sA failed 0x%x\n", ret ); > + ok( ret==0x55 || ret == 0x51, "ldap_search_ext_sA unexpected return 0x%x\n", ret ); This should be fixed by my last patch already. -Hans

2 1

Re: wintrust: Implement CryptCATAdminCalcHashFromFileHandle, try 2
by Alexandre Julliard 21 Oct '08

21 Oct '08

Maarten Lankhorst <m.b.lankhorst(a)gmail.com> writes: > + curpos.QuadPart = 0; > + SetFilePointerEx(hFile, curpos, &oldpos, FILE_CURRENT); > + SetFilePointerEx(hFile, curpos, NULL, FILE_BEGIN); That's not thread-safe. -- Alexandre Julliard julliard(a)winehq.org

2 1