Age | Commit message (Collapse) | Author |
|
CID 1351617, 1351618, 1351619, 1351620, 1351621, 1351622, 1354791.
|
|
This ensures that if there is a bug in the drawing code that causes
the row to be unset, it will not result in silently reading garbage
out of random memory.
CID 1354802
|
|
The previous implementation did not work properly, assigning the
next insertion index to the oldest object instead of filling empty
cache slots on a cache miss.
|
|
|
|
|
|
|
|
|
|
The priority map is not used at all in SCI32 at the engine level by
design, so all the relevant code that handles picture priority is
pretty much dead
|
|
Still not working
|
|
In order for scaling ratios to apply equally across objects that
start at different positions on the screen, the pixels that are
read from the source bitmap must all use the same pattern of
division. In other words, cels must follow the same scaling pattern
as if they were drawn starting at an even multiple of the scaling
ratio, even if they were not.
|
|
This reverts commit d85eb8ded68a20de383d84064aacd1a4c81db4e9.
This patch did not correctly fix the scaler to follow the same
rules as SSCI and only worked on the y-axis.
|
|
Previously sourcePos was always originating from plain 0, 0
which made some pixels not always getting drawn at the right
spot when uneven scaling was used (for example 5:12).
Seems to fix gabriel knight 1 hires graphic issues
|
|
|
|
|
|
|
|
|
|
Not all SCI2.1late/SCI3 function signatures are fully known yet,
but all subops are now represented in the kernel tables.
|
|
|
|
1. Reorder member initialisations to match class member order
2. Use #pragma mark instead of comments for annotating sections
3. Remove useless >=0 checks on unsigned types
|
|
This replaces "colour" to "color"
|
|
|
|
|
|
|