[wined3d] support for WINED3DFMT_A8B8G8R8

[Christoph Frick]

attempt #3:
this patch fixes several texture bugs with the demos of OGRE.

License: LGPL
ChangeLog:
2006-02-08  Christoph Frick 
    - dlls/wined3d/utils.c
    support for WINED3DFMT_A8B8G8R8

[Karsten Elfenbein]

The patch looks more like repeated code to me.

FMT_TO_STR(WINED3DFMT_A8B8G8R8);
then add a second
FMT_TO_STR(WINED3DFMT_A8B8G8R8);

Doesn't make sense for me. Same for the other changes in this diff.

Karsten
> -----Ursprüngliche Nachricht-----
> Von: [e-mail] 
> [mailto:[e-mail]] Im Auftrag von Christoph Frick
> Gesendet: Mittwoch, 8. Februar 2006 13:32
> An: [e-mail]; [e-mail]
> Betreff: [wined3d] support for WINED3DFMT_A8B8G8R8
> Wichtigkeit: Hoch
>
> attempt #3:
> this patch fixes several texture bugs with the demos of OGRE.
>
> License: LGPL
> ChangeLog:
> 2006-02-08  Christoph Frick
> - dlls/wined3d/utils.c
>
> support for WINED3DFMT_A8B8G8R8
>
> -- 
> cu
> Index: dlls/wined3d/utils.c
>
> RCS file: /home/wine/wine/dlls/wined3d/utils.c,v
> retrieving revision 1.18
> diff -u -r1.18 utils.c
> --- dlls/wined3d/utils.c	23 Nov 2005 19:29:20 -0000	1.18
> +++ dlls/wined3d/utils.c	8 Feb 2006 12:23:00 -0000
> @@ -35,6 +35,7 @@
>      FMT_TO_STR(WINED3DFMT_UNKNOWN);
>      FMT_TO_STR(WINED3DFMT_R8G8B8);
>      FMT_TO_STR(WINED3DFMT_A8R8G8B8);
> - FMT_TO_STR(WINED3DFMT_A8B8G8R8);
> FMT_TO_STR(WINED3DFMT_X8R8G8B8);
> FMT_TO_STR(WINED3DFMT_R5G6B5);
> FMT_TO_STR(WINED3DFMT_X1R5G5B5);
>
> @@ -1611,6 +1612,7 @@
>          case WINED3DFMT_A4R4G4B4:         retVal = GL_RGBA4; break;
>          case WINED3DFMT_X4R4G4B4:         retVal = GL_RGB4; break;
>          case WINED3DFMT_A8R8G8B8:         retVal = GL_RGBA8; break;
> -
>    case WINED3DFMT_A8B8G8R8:         retVal = GL_RGBA8; break;
>    case WINED3DFMT_A2R10G10B10:      retVal = GL_RGBA8; break;
>    case WINED3DFMT_X8R8G8B8:         retVal = GL_RGB; break;
>        /* to see */
>
> @@ -1676,6 +1678,7 @@
>          case WINED3DFMT_A4R4G4B4:         retVal = GL_BGRA; break;
>          case WINED3DFMT_X4R4G4B4:         retVal = GL_BGRA; break;
>          case WINED3DFMT_A8R8G8B8:         retVal = GL_BGRA; break;
> -
>    case WINED3DFMT_A8B8G8R8:         retVal = GL_RGBA; break;
>    case WINED3DFMT_A2R10G10B10:      retVal = GL_BGRA; break;
>    case WINED3DFMT_X8R8G8B8:         retVal = GL_BGRA; break;
>        /* to see */
>
> @@ -1738,6 +1741,7 @@
>          case WINED3DFMT_A4R4G4B4:         retVal = 
> GL_UNSIGNED_SHORT_4_4_4_4_REV; break;
>          case WINED3DFMT_X4R4G4B4:         retVal = 
> GL_UNSIGNED_SHORT_4_4_4_4_REV; break;
>          case WINED3DFMT_A8R8G8B8:         retVal = 
> GL_UNSIGNED_INT_8_8_8_8_REV; break;
> -
>    case WINED3DFMT_A8B8G8R8:         retVal = 
>
> GL_UNSIGNED_INT_8_8_8_8_REV; break;
>          case WINED3DFMT_A2R10G10B10:      retVal = 
> GL_UNSIGNED_INT_2_10_10_10_REV; break;
>          case WINED3DFMT_X8R8G8B8:         retVal = 
> GL_UNSIGNED_INT_8_8_8_8_REV; break;
>              /* to see */
> @@ -1781,6 +1785,7 @@
>      case WINED3DFMT_A4R4G4B4:         retVal = 2; break;
>      case WINED3DFMT_X4R4G4B4:         retVal = 2; break;
>      case WINED3DFMT_A8R8G8B8:         retVal = 4; break;
> - case WINED3DFMT_A8B8G8R8:         retVal = 4; break;
> case WINED3DFMT_A2R10G10B10:      retVal = 4; break;
> case WINED3DFMT_X8R8G8B8:         retVal = 4; break;
>     /* Paletted */

[H. Verbeet]

On 08/02/06, Karsten Elfenbein  wrote:
> The patch looks more like repeated code to me.
>
> FMT_TO_STR(WINED3DFMT_A8B8G8R8);
> then add a second
> FMT_TO_STR(WINED3DFMT_A8B8G8R8);
>
> Doesn't make sense for me. Same for the other changes in this diff.
>
> Karsten

Unless I'm misunderstanding your comment, that's not what the patch does.
There are already cases for WINED3DFMT_A8R8G8B8, the patch adds cases
for WINED3DFMT_A8B8G8R8. I suppose the R and the B look rather
similar.

[Christoph Frick]

On Wed, Feb 08, 2006 at 08:11:47PM -0400, H. Verbeet wrote:
> > FMT_TO_STR(WINED3DFMT_A8B8G8R8);
> > then add a second
> > FMT_TO_STR(WINED3DFMT_A8B8G8R8);
> > Doesn't make sense for me. Same for the other changes in this diff.
>
> Unless I'm misunderstanding your comment, that's not what the patch
> does.  There are already cases for WINED3DFMT_A8R8G8B8, the patch adds
> cases for WINED3DFMT_A8B8G8R8. I suppose the R and the B look rather
> similar.

even without doing any further explanation: this patch cant be
duplication because gcc handles duplicate case values as an error.

but you are right - thats exactly what it does (note the reversed order:
RGB vs BGR). without this patch there are fallbacks for RGB, 4byte, and
so on - so there is no actual breakage code-wise. BUT without this patch
running the ``smoke'' and ``grass'' demo show heavily distorted
textures. the ogre-logo used in every demo is distorted too. this patch
fix this behaviour. i doubt that many other games have similar problems,
because ogre is designed to work with ogl and d3d and would therefore
deal with textures in both flavours and the demos may not be written by
the same author and the textures may not all be of the same format or
source.