| Re: [AD] About Elias' bug |
[ Thread Index |
Date Index
| More lists.liballeg.org/allegro-developers Archives
]
> (I ran the complete pink.exe for both diffs at
> http://www.allefant.f2s.com/allegro/timings.html)
Wow !
However, I'm a little puzzled by the results you got for conversions from
8-bit (8->15 normal specifically), given that the codes in blit3b.diff and
blit4.diff are exactly the same.
Anyway, I'm going to commit the blit4.diff patch.
> AFAICS all conversions work correctly. I see a /* to do */ at the 8bit
> replacement color, and I didn't try long enough to understand why it works
> with 0, but it works in the test program :)
It works only if you set up a RGB mapping table, because such tables never
contain index 0. Otherwise, it doesn't work because best_fit_color() does
return 0 for the pink color.
But I'm waiting until we decide what to do with index 0 in paletted modes:
if we keep it as a special index, I will make best_fit_color() never return
0 and get_replacement_color() won't have any more thing to do; otherwise, it
will have to use a tweaked version of best_fit_color().
--
Eric Botcazou
ebotcazou@xxxxxxxxxx