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

Re: try2:patch for sysparams.c

by L. Rahyen

On Wednesday August 29 2007 09:15, Ivan Sinitsin wrote: > Has corrected "logpixels=72" to " logpixels = 96 " and has changed the > indentation > > This patch adds an opportunity for fonts of the menu, statusbar and > messages to increase depending on logpixels. > > > Changelog: > This patch does a font for the menu,statusbar and messages dependent from > logpixels. With your patch applied to current GIT head I have too large fonts in all menus even in notepad if for example I have set logpixels to 120 dpi using winecfg. "Too large" means MUCH larger than in, for example, winecfg itself (you need to restart the program to see the effect). See atached snapshot to see how notepad looks like with your patch if I set 120 dpi in winecfg. Font size is ridiculously big! For testing I have used clean WINE with your patch only and clean WINEPREFIX. In Windows and current WINE without your patch in clean WINEPREFIX all fonts are of same size no matter what DPI I set in LogPixels or winecfg and this is correct behavior. What application you are trying to fix?

Aug. 29, 2007

Re: [2/5] WineD3D: Clear unused channels on R32F and R16F textures on readback

by H. Verbeet

On 29/08/2007, Stefan Dösinger <stefan(a)codeweavers.com> wrote: > Can you even use those formats as render target on any cards?

Aug. 29, 2007

Re: winex11.drv[2/2]: Support GdiAlphaBlend with blendfn.AlphaFormat == 0 (fixes bug #9262)

by Alexandre Julliard

"Mikolaj Zalewski" <mikolajz(a)google.com> writes: > @@ -1600,6 +1603,16 @@ #endif > for(; y >= y2; y--) { > memcpy(dstbits, (char *)dib.dsBm.bmBits + y * dib.dsBm.bmWidthBytes + xSrc * 4, > widthSrc * 4); > + if (!(blendfn.AlphaFormat & AC_SRC_ALPHA)) > + { > + int i; > + RGBQUAD *quad = (RGBQUAD *)dstbits; > + for (i = 0; i < widthSrc; i++) > + { > + quad->rgbReserved = blendfn.SourceConstantAlpha; > + quad++; > + } > + } It would be nicer to have a separate loop for that case, and do everything in one go instead of patching up the values after the copy. Also RGBQUAD is meant for color tables, it's not an appropriate type for the bitmap data itself. -- Alexandre Julliard julliard(a)winehq.org

Aug. 29, 2007

Re: user32: static controls should have a clipping region set while sending the WM_CTLCOLORSTATIC (3rd try, with testcase, fixes bug #9195)

by Alexandre Julliard

"Mikolaj Zalewski" <mikolajz(a)google.com> writes: > Is there something wrong with this patch? You should use IntersectClipRect or ExtSelectClipRgn to avoid intermediate copies. Also GetClipRgn doesn't return a region code so checking it against NULLREGION is wrong. -- Alexandre Julliard julliard(a)winehq.org

Aug. 29, 2007

Re: patch for sysparam.c

by Paul Vriens

Ivan Sinitsin wrote: > This patch adds an opportunity for fonts of the menu, statusbar and messages > to increase depending on logpixels. > > > Changelog: > This patch does a font for the menu,statusbar and messages dependent from > logpixels. > > > ------------------------------------------------------------------------ > > With the latest release winecfg can set this value. The default there is 96 and the range is 96..120. How does that relate to the 72 in your code? : +if (!RegQueryValueExW( myhKey, log_pixels, NULL, &type, + (LPBYTE) &logpixels, &size )==ERROR_SUCCESS) + logpixels=72; Cheers, Paul.

Aug. 29, 2007

Re: patch for sysparam.c

by Robert Shearman

Ivan Sinitsin wrote: > This patch adds an opportunity for fonts of the menu, statusbar and messages > to increase depending on logpixels. > > > +size=sizeof(DWORD); > +RegOpenKeyExW( HKEY_LOCAL_MACHINE, registry_key, > + 0, KEY_QUERY_VALUE, &myhKey ); > +if (!RegQueryValueExW( myhKey, log_pixels, NULL, &type, > + (LPBYTE) &logpixels, &size )==ERROR_SUCCESS) > + logpixels=72; > +ncm.lfMenuFont.lfHeight=-MulDiv(ncm.lfMenuFont.lfHeight,logpixels,72); > +RegCloseKey( myhKey ); You're missing some indentation here. -- Rob Shearman

Aug. 29, 2007

Re: [PATCH] Support for animated cursors

by Trent Waddington

On 8/29/07, Juan Lang <juan.lang(a)gmail.com> wrote: > This doesn't belong with this patch. Yes, sorry. > One more random piece of advice: this has gotten enough commentary > that it may very well have entered Alexandre's "ignore for a while" > queue. So if you don't have success with this, wait some time, and > try again. Ok, no problem. Final(?) patch attached. Trent

Aug. 28, 2007

Re: [PATCH] Support for animated cursors

by Juan Lang

--- a/dlls/imm32/imm.c +++ b/dlls/imm32/imm.c @@ -1505,7 +1505,8 @@ BOOL WINAPI ImmRegisterWordW( */ BOOL WINAPI ImmReleaseContext(HWND hWnd, HIMC hIMC) { - FIXME("(%p, %p): stub\n", hWnd, hIMC); + if (hIMC == (HIMC)root_context) + root_context->IMC.hWnd = NULL; This doesn't belong with this patch. > if you pop into the wine/dlls/user32 dir and do a grep 0x00030000 > * you'll see I'm not the first to follow the documentation. Fair enough, objection withdrawn. One more random piece of advice: this has gotten enough commentary that it may very well have entered Alexandre's "ignore for a while" queue. So if you don't have success with this, wait some time, and try again. --Juan

Aug. 28, 2007

Re: [PATCH] Support for animated cursors

by Trent Waddington

On 8/29/07, Juan Lang <juan.lang(a)gmail.com> wrote: > That doesn't make it better, it's not thread-safe. Pass them as > parameters to decodeRIFF. Ahh, I see. Yes, fixed. > I still say a symbolic constant is better, or at the least, a comment. > A test case that shows that it's needed is even better. http://msdn2.microsoft.com/en-us/library/ms648061.aspx dwVersion [in] Specifies the version number of the icon or cursor format for the resource bits pointed to by the pbIconBits parameter. This parameter can be 0x00030000. and if you pop into the wine/dlls/user32 dir and do a grep 0x00030000 * you'll see I'm not the first to follow the documentation. It's cryptic, but hey, it's Microsoft. Trent

Aug. 28, 2007

Re: [PATCH] Support for animated cursors

by Juan Lang

> > Like Henri already said, you really can't use globals for anih, > > ani_frames, and ani_frame_idx. These should be in a structure, and > > decodeRIFF should write into that structure, not into globals. The > > structure should then be associated with the icon. > > They're temps.. I've added a comment to make that clear. That doesn't make it better, it's not thread-safe. Pass them as parameters to decodeRIFF. > > This may be obvious to other, but certainly not to me: what does > > 0x00030000 mean? You need to use symbolic constants here. > > It's the version non-sense that Microsoft's api requires. You'll find > that every call to this api and it's non-Ex variant have it. I still say a symbolic constant is better, or at the least, a comment. A test case that shows that it's needed is even better. --Juan

Aug. 28, 2007