Re: [AD] Patch for two new entries to GFX_VTABLE

[Bob <ohannessian@xxxxxxxxxx>]

Eric Botcazou wrote:

> --- src/blit.c 2001/11/06 15:30:46 1.9
> +++ src/blit.c 2001/11/08 03:57:29
> @@ -785,9 +785,11 @@
>  {
>     BLIT_CLIP();
>
> -   if (!(is_video_bitmap(dest) && is_video_bitmap(src)) &&
> -       (src->vtable->color_depth != dest->vtable->color_depth)) {
> +   if (src->vtable->color_depth != dest->vtable->color_depth) {
>        /* need to do a color conversion */
> +      if (dest->vtable->blit_between_formats)
> +         dest->vtable->blit_between_formats(src, dest, s_x, s_y, d_x, d_y,
> w, h);
> +      else
>        blit_between_formats(src, dest, s_x, s_y, d_x, d_y, w, h);
>     }
>     else if (is_same_bitmap(src, dest)) {
>
>
> Why did you remove the (!(is_video_bitmap(dest) && is_video_bitmap(src))
> condition ? The whole if/then/else chain is modified.


Not really, this just undoes an old patch I submitted a while ago, which was 
a hack for AllegroGL. The condition up there should never be met in regular 
Allegro programs - you can't have video bitmaps of a different color depth 
than the screen bitmap. However, in AllegroGL you can, so my previous patch 
inhibited the color conversion for that case. Now that we have a vtable 
entry for converting in between color formats, I can easily do everything 
needed from within AllegroGL.


> I suppose you want to do your own color conversion for video bitmaps only.


Like I said, that can't happen ;)
If you want to do memory bitmap -> video bitmap conversion (or inversely), 
then nothing has changed - the regular blit_between_formats gets called.

> Why not rather hook the blit_to_self() method of the vtable ?


Because video bitmaps are not necessairly part of the screen (AllegroGL) ;)

[snip]


> This one looks ok, except that since the new method is a replacement for
> _pivot_scaled_sprite_flip() in rotate.c, it should be invoked wherever
> _pivot_scaled_sprite_flip() is invoked, not inside
> _pivot_scaled_sprite_flip() itself.


Ok, I'll submit a replacement patch.



--
- Robert J Ohannessian
"Microsoft code is probably O(n^20)" (my CS prof)
http://pages.infinit.net/voidstar/