The PS2 versions are not working at all. Furthermore,
the detection entries caused AdvancedMetaEngine to output
a misleading error message when the game executable cannot
be found.
Both the animation components/tracks and the chores itself keep track of
the current time independently:
animation component:
- AnimationStateEmi::_time
- initialized with 0
- after the first update() call, _time is advanced by the frametime
chores:
- Costume::_currTime
- initialized with -1
- during the first update() call, _currTime is set first to 0
- during the next update() call, _currTime is advanced by the frametime
EMICostume::update() updates first the chore and then all components
with the same frame time. This has the effect, that the internal time
base of the chore and the animation has always an initial offset (the
value of the frametime of the first update() call).
For some costumes, the chore will pause the animation by setting the
according key for that animation at the end of the animation. However,
since the animation component has a different time base, it has already
reached that point and may have already disabled itself entirely
(instead of just being paused by the chore).
This bug fix let the animation set _time to 0 in the first update() call, too.
Any consecutive update() call will advance the time by the frametime.
This will keep the two time bases of the chore and the animation component
in sync.
Instead of just ordering the textobjects by their layer, use their
creation time (indirectly provided by the pool ID which is continuously
counting up).
This ensures that if there are multiple textobjects with the same text
ID and the same layer, that the latest is drawn last and overwrites the
previous ones.
That's necessary since the LUA code will create multiple instances of
the same textobject for the color fading effect in the credits.
This patch enables drawing of Layer objects which are
dynamically created by LUA code.
All three sorted lists (background images, layers and actors) are
merged based on their sort-order.
Instead of using the number of sub-images to decide whether bitmaps
should be drawn using TIL information, better use the existance
of texture coordinates which are distinctive for tiles.
When switching between shaders, the previous shader might have Y
attributes, while the current shader only has X, with X < Y. The extra
attributes will not be used by the shader, but the driver might attempt
to read from the bound FBOs anyway. OSX's OpenGL profiler stumbles over
this.
