this laregly occured due to adding ribbon nodes in smaller increments than the size of a given quad.
we therefore skip adding new ribbon nodes if velocity, as determined by the point deltas, would be below 10% of the largest scale a given quad for that link in the ribbon would be.
as this will result in lower than a max length ribbon, also adds a timeout mechanism of removing a link every TickMS (32 ticks/second, or roughly how often one would also be aded when in a growth state)
for full finalization do still need to circle back and adress why there remains 1 quad spawned after motion.
- Clicking the [...] button from the Particle Emitter tab now opens to the Particle tab *and* properly selects the particle data to be edited
- Selecting an emitter object in the map and opening the particle editor now selects the Particle Data to be edited
- Selecting a new Particle Data for a particles slot on a Particle Emitter in the editor now correctly updates the values and updates the field display
- Made it so if clicking [...] button on the ParticleEmitterData or other similar fields for objects, it will now open to the Particle editor instead of the Datablock editor
add minimal padding to counteract floating point flurrer when rounding due to changes in parent element scale
revisit the aspect maintining resizer math to correct a few failures with live editing the values.
the player class (and resource derivatives, contain a mix of hardcoded animation names, as well as scriptable ones that can be tripped via playthread/setActionThread.
to determine if an animation within the stored vector is hardcoded or a scripted oneoff for that mesh, theres a demarcation at NumTableActionAnims for the up to 512 animation slots available.
when deriving from player, we must therefore allow that entry to be overridden for any class which adds additional hardcoded animations
therefore this introduces a datablock-level entry for the marked slot.
update bitmapPng reading from libpng to have context in warnings and errors, also add a bit of a better structure around the read and write functions.
Interlaced PNG's were never being accounted for before so add the png_set_interlace_handling for interlaced pngs
clearing the active program caused issues during a target bake as we call glClear after setting the shader etc. As this warning id 131204 seems to be a product of doing this, everything i read up on it says it is safe to silence it.
The core of this issue was the unlock on the GFXGLTextureObject
Few other bug fixes around setting a texture binding to 0, notice in clear function we now also clear the glprogram and the buffers. This seems to fix most of the warnings around id 131204
This warning was triggered every frame, now it just triggers when a shader expects a texture but none is active.
player and vehicle were not loading some of thier sounds.
we'll want to revisit this with a more comprehensive solution in the future, but this will at least allow folks to use the systems in place
There is an exception on teardown usually during a debug build that is caused during shapeBase onRemove calling hasAccumulation.
This fix reorders the hasAccumulation check to be before shapebase calls Parent::onRemove, and also hard sets the mMaterialList to NULL when setMaterialList is called
from marauder:
Fixing transform plane as some descrepencies between the simd function and the c math caused issues with precision.
Use c math for inverse functions until a proper simd implementation can be found.
In Player::updatePos(), when processing physics collision results, the code was incorrectly using col.object instead of colCheck.object when checking if the collision object is a player.
This caused:
- First iteration: col.object is uninitialized (zeroed), leading to null pointer access
- Subsequent iterations: col.object contains the previous iteration's value, causing incorrect type checks
The fix changes col.object to colCheck.object to properly check the current collision object.
the opened ddsfs file for pulling metadata was leaking. The FileStream function close does not delete the pointer to the original file only FileObject does this. It might be worthwhile instead to switch to using fileObject anywhere we use a filestream that way we can be at least somewhat safe in using the object and once it goes out of scope the raw pointer will be freed
