- Wine-Devel - list.winehq.org

wine source doubts
by Preetham July 9, 2002

July 9, 2002

Hi All: I am a newbie to wine. I have a few queries. 1. I downloaded the wine source from cvs, however, is there any document as to where i can start lookin into the code.? 2. Where is the program(wine loader) that loads a .ece file located in the source. Appriciate the help. Regards, Preetham

2 1

One more try at bug 829 (same surface corruption)
by Tony Lambregts July 9, 2002

July 9, 2002

After talking with Lionel Ulmer offline I came up with the following patch for bug 829. It includes a hopefully more clear comment of when the corruption can occure and a better set of rules. Change log: fix same surface corruption when no flags are set and areas are the same size Tony Lambregts tony_lambregts(a)telusplanet.net Index: dib.c =================================================================== RCS file: /home/wine/wine/dlls/ddraw/dsurface/dib.c,v retrieving revision 1.11 diff -u -r1.11 dib.c --- dib.c 28 Jun 2002 17:32:25 -0000 1.11 +++ dib.c 9 Jul 2002 03:57:37 -0000 @@ -351,6 +351,7 @@ int bpp, srcheight, srcwidth, dstheight, dstwidth, width; int x, y; LPBYTE dbuf, sbuf; + BOOL SameSurfaceOK = TRUE; TRACE("(%p)->(%p,%p,%p,%08lx,%p)\n", This,rdst,src,rsrc,dwFlags,lpbltfx); @@ -474,6 +475,18 @@ sbase = (BYTE*)sdesc.lpSurface+(xsrc.top*sdesc.u1.lPitch)+xsrc.left*bpp; xinc = (srcwidth << 16) / dstwidth; yinc = (srcheight << 16) / dstheight; + SameSurfaceOK = ((src != iface)|| + (sbase >= dbuf)|| + (xdst.top > xsrc.bottom)||(xdst.left > xsrc.right)||(xdst.right < xsrc.left)); + + /* + A little explination: Screen coruption can occure if the following set of circumstances is true + + 1.) Both the source and destination are the same surface. -->(src != iface) + 2.) The start of the source is less than the destination . -->(sbase >= dbuf) + 3.) There is overlap. -->(xdst.top < xsrc.bottom)&&(xdst.left < xsrc.right)(xdst.right < xsrc.left) + This is fixed in the case where no flags are set and the areas are the same size. + */ if (!dwFlags) { /* No effects, we can cheat here */ @@ -482,11 +495,29 @@ /* No stretching in either direction. This needs to be as * fast as possible */ sbuf = sbase; - for (y = 0; y < dstheight; y++) { - memcpy(dbuf, sbuf, width); - sbuf += sdesc.u1.lPitch; - dbuf += ddesc.u1.lPitch; - } + if (SameSurfaceOK) { + for (y = 0; y < dstheight; y++) { + memcpy(dbuf, sbuf, width); + sbuf += sdesc.u1.lPitch; + dbuf += ddesc.u1.lPitch; + } + } else { + sbuf += (sdesc.u1.lPitch*dstheight); + dbuf += (ddesc.u1.lPitch*dstheight); + if (xdst.top == xsrc.top) { /* Are the source and destination on the same line? */ + for (y = 0; y < dstheight; y++) { + sbuf -= sdesc.u1.lPitch; + dbuf -= ddesc.u1.lPitch; + memmove(dbuf, sbuf, width); + } + } else { + for (y = 0; y < dstheight; y++) { + sbuf -= sdesc.u1.lPitch; + dbuf -= ddesc.u1.lPitch; + memcpy(dbuf, sbuf, width); + } + } + } } else { /* Stretching in Y direction only */ for (y = sy = 0; y < dstheight; y++, sy += yinc) { @@ -544,6 +575,10 @@ } } else if (dwFlags & (DDBLT_KEYSRC | DDBLT_KEYDEST | DDBLT_KEYSRCOVERRIDE | DDBLT_KEYDESTOVERRIDE)) { DWORD keylow, keyhigh; + + if (!SameSurfaceOK) { + FIXME("\tSoure and destination surfaces are the same and could cause display problems\n"); + } if (dwFlags & DDBLT_KEYSRC) { keylow = sdesc.ddckCKSrcBlt.dwColorSpaceLowValue;

1 0

Re: SHLWAPI wsprintf - with patch this time
by Alexandre Julliard July 9, 2002

July 9, 2002

Jon Griffiths <jon_p_griffiths(a)yahoo.com> writes: > +dlls/shlwapi/wsprintf.c dlls/user/wsprintf.c > Add decimal double (%f) format specifier to *printf functions. These functions are not supposed to handle %f. If you need %f you have to use the real sprintf, not the wsprintf variants. -- Alexandre Julliard julliard(a)winehq.com

1 0

malloc, heapalloc, etc
by John Lightsey July 8, 2002

July 8, 2002

Hi all, I've been playing around with the wininet dll to get a windows app I developed running, and I've written a patch that fixes the problems with wininet that I've noticed, but it's an ugly fix and I'd like to get a little clarification about how heap memory is being used in Wine. In dlls/wininet/internet.c InternetOpenUrlA() malloc() is used to put together the HTTP host header in hostreq. The pointer is then passed to HttpAddRequestHeadersA() which creates a buffer with HeapAlloc, copies the contents pointed to by hostreq to it then frees the buffer with HeapFree. Control then passes back to InternetOpenUrlA which does not call free() on the hostreq pointer. So, my question is which is the encouraged way of doing things? HeapAlloc and HeapFree or malloc and free? Also, is the fact that free() was not called on hostreq a bug, or am I missing something? Please be gentle....I'm a Perl programmer, not a C programmer.

2 1

about programming books
by Andreas Mohr July 8, 2002

July 8, 2002

Hi all, I just created http://wine.codeweavers.com/fom-meta/cache/831.html to list all the Windows programming books I own (16 books ! ;-) I'd be happy if some other people could add an entry, too. That way we'd have an amazing information resource whenever needed. Hmm, I just hope you've got editing capabilities on this page. If you're unable to add something, then please tell me. (or just edit an entry in some FOM category where you're able to edit it, I'll move it there then) Of course I will add more details of the books to the page once I feel bored (actually a separate page per book within a sub category per author would probably be better). Thanks ! -- Microsoft Licensing 6.0: "Pay us now in advance, so that we can own you later."

1 0

Support for 24bpp when DDBLT_KEYSRC etc are set
by Tony Lambregts July 8, 2002

July 8, 2002

I am working on implementing the flags under DDBLTFX (Bug 841) and in order to do so I needed to understand how the current code worked. While researching this I came up with this patch for the flags DDBLT_KEYSRC/DDBLT_KEYDEST and company. This patch should be faster then the current code since the switch statement is outside the loop. It also adds support for 24bpp colour. The problem that I have with this is that it assumes stretching in both the x and y directions but from what I understand stretching should only be done if the flag DDBLTFX is set and has the operation DDBLTFX_ARITHSTRETCHY in dwDDFX. Is that right or can Marcus Meissner or Lionel Ulmer explain this to me. Change log: added support for 24bpp when the flags DDBLT_KEYSRC/DDBLT_KEYDEST and company are set. Tony Lambregts tony_lambregts(a)telusplanet.net Index: dib.c =================================================================== RCS file: /home/wine/wine/dlls/ddraw/dsurface/dib.c,v retrieving revision 1.11 diff -u -r1.11 dib.c --- dib.c 28 Jun 2002 17:32:25 -0000 1.11 +++ dib.c 8 Jul 2002 19:02:35 -0000 @@ -561,28 +561,54 @@ keyhigh = lpbltfx->ddckDestColorkey.dwColorSpaceHighValue; } - for (y = sy = 0; y < dstheight; y++, sy += yinc) { - sbuf = sbase + (sy >> 16) * sdesc.u1.lPitch; #define COPYROW_COLORKEY(type) { \ - type *s = (type *) sbuf, *d = (type *) dbuf, tmp; \ - for (x = sx = 0; x < dstwidth; x++, sx += xinc) { \ - tmp = s[sx >> 16]; \ - if (tmp < keylow || tmp > keyhigh) d[x] = tmp; \ - } \ - break; } + type *s = (type *) sbuf, *d = (type *) dbuf, tmp; \ + for (y = sy = 0; y < dstheight; y++, sy += yinc) { \ + (LPBYTE)s = sbase + (sy >> 16) * sdesc.u1.lPitch; \ + for (x = sx = 0; x < dstwidth; x++, sx += xinc) { \ + tmp = s[sx >> 16]; \ + if (tmp < keylow || tmp > keyhigh) d[x] = tmp; \ + } \ + (LPBYTE)d += ddesc.u1.lPitch; \ + } \ + break; } - switch (bpp) { - case 1: COPYROW_COLORKEY(BYTE) - case 2: COPYROW_COLORKEY(WORD) - case 4: COPYROW_COLORKEY(DWORD) - default: - FIXME("%s color-keyed blit not implemented for bpp %d!\n", - (dwFlags & DDBLT_KEYSRC) ? "Source" : "Destination", bpp*8); - ret = DDERR_UNSUPPORTED; - goto error; - } - dbuf += ddesc.u1.lPitch; + switch (bpp) { + case 1: COPYROW_COLORKEY(BYTE) + case 2: COPYROW_COLORKEY(WORD) + case 4: COPYROW_COLORKEY(DWORD) + case 3: { + int last_sy = -1; + for (y = sy = 0; y < dstheight; y++, sy += yinc) { + sbuf = sbase + (sy >> 16) * sdesc.u1.lPitch; + if ((sy >> 16) == (last_sy >> 16)) { + /* this sourcerow is the same as last sourcerow - + * copy already stretched row + */ + memcpy(dbuf, dbuf - ddesc.u1.lPitch, width); + } else { + LPBYTE s,d = dbuf; + for (x = sx = 0; x < dstwidth; x++, sx+= xinc) { + DWORD pixel; + + s = sbuf+3*(sx>>16); + pixel = (s[0]<<16)|(s[1]<<8)|s[2]; + d[0] = (pixel>>16)&0xff; + d[1] = (pixel>> 8)&0xff; + d[2] = (pixel )&0xff; + d+=3; + } + } + dbuf += ddesc.u1.lPitch; + last_sy = sy; + } + break;} + default: + FIXME("%s color-keyed blit not implemented for bpp %d!\n", + (dwFlags & DDBLT_KEYSRC) ? "Source" : "Destination", bpp*8); + ret = DDERR_UNSUPPORTED; + goto error; } #undef COPYROW_COLORKEY dwFlags &= ~(DDBLT_KEYSRC | DDBLT_KEYDEST | DDBLT_KEYSRCOVERRIDE | DDBLT_KEYDESTOVERRIDE);

1 0

RE: Wine ODBC compatibilty
by Medland, Bill July 8, 2002

July 8, 2002

> -----Original Message----- > From: James Juran [mailto:jamesjuran(a)alumni.psu.edu] > Sent: Friday, June 28, 2002 3:43 PM > To: Robert Damen > Cc: 'wine-devel(a)winehq.com' > Subject: Re: Wine ODBC compatibilty > > > Robert Damen wrote: > > > > Wine, > > I found this address on your website, if it is the > wrong address for > > these types of questions - Sorry!. > > > > I was wondering weather you could answer a question I have > about support for > > Microsoft Office applications. > > The question is about weather wine can emulate (or has its > own version of) > > the "ODBC Data Sources (32bit)" application found in > > c:\windows\Control Panel\ODBC Data Sources > > This application is used to set up DSN drivers for ODBC > access on Access and > > Excel. > > An example of one of the drivers that this application > handles is the MyODBC > > 3.51 windows driver. > > The "ODBC Data Sources" application in the control panel merely > manipulates registry entries. If all you need to do is set > them up, you > can do that easily with regedit. > > If you're wondering if they will work after you set them up, someone > else will have to answer that question :-) > > -- > James Juran > jamesjuran(a)alumni.psu.edu > To the best of my knowledge there is no such application, and it would be strange if there were. The WINE ODBC32 module is merely a wrapper around any suitable Linux ODBC module. For example we use unixODBC. Thus the WINE ODBC module has no real need to access the registry since all the work is being done by the Linux ODBC module. The way that the Linux ODBC module is configure depends upon which one you use. The unixODBC module, for example, uses a odbc.ini and odbcinst.ini file to store its configuration information. Although it does have a graphical configuration tool, with plugins for some database drivers, I prefer to edit the files by hand. That being said, it may be necessary to add some information to the registry for some ODBC-based applications. I do know that Crystal Reports directly accesses the registry for ODBC information rather than asking ODBC itself. HTH Bill

1 0

Re: Question about --debugmsg +snoop
by Massimo July 8, 2002

July 8, 2002

So if I understand well there's no way now to snoop a single DLL, even putting it in ~/.wine/config ?

1 0

Borland linker and VirtualQuery
by Vassilis Virvilis July 8, 2002

July 8, 2002

Halo, The attached patch (resurrected from) wine-users through google searching fixes indeed the borland command line tools (version 5.5.1 freely downloadable on the internet). Given that I am not familiar with the win32 API, this patch looks hackish and special cased (admitted also by its author, Gerard Patel) but still it is too useful. Since this question has never surfaced to wine-devel I would like to ask what is required for the patch to become acceptable. If it is something doable I will give it a try, otherwise I will file a report in the wine bugzilla database. Any way thanks in adavance. .Bill

1 0

Question about --debugmsg +snoop
by Massimo July 8, 2002

July 8, 2002

Is there a way to activate +snoop only for selected dll ??? I think that would help me to find AutoCAD 2000 access violation on startup, isolating the dll where it happens. As stated few msg before, it starts up good with +snoop, but it crashes on very beginning without +snoop. Regards Max

3 4