also change the fullVertices to the same format
change the way the coordinates of stickers are kept from float[2*N] to float[N][2]
minor
Major progress with supporting cubit faces with holes.
Initial support for scrambling a bandaged pyraminx.
Improve MeshMultigon (for the BandagedPyraminx case!)
another bugfix for MeshMultigon.
bugfix for a crasher
Bugfix in MeshMultigon.
remove commented out code
We no longer need to support 'edgesUp' in MeshPolygons.
Fixes for MeshMultigon. Perfect!
Implement MeshMultigon out of BandedTriangles (doesn't work yet)
MeshBandedTriangle implemented
MeshBandedTriangle
bugfix for a bugfix :)
Fix a bug noticed by Rohan M.
Comment.
correct face shapes (post-new MeshPolygon normal computation)
improve MeshPolygon. Now Multigon's meshes are close to perfect.
Multigon testing app
MeshMultigon finished.
Progress with MeshMultigon. Still auto-computation of 'up' edges does not work.
Beginnings of MeshMultigon (does not work yet)
MeshPolygon: support for an arbitrary subset of edges being 'up' (i.e. buildup to MeshMultigon)
MeshPolygon: fix the normals.
one more fix for displaying normals.
Fix bug which prevented the normal vectors from being drawn.Bug introduced in 2021/Dec/16!
build for API 34.
speedup
Completely kick out the android.opengl.Matrix dependency from the library.
Implement the android.opengl.Matrix functions ourselves.
Make the touchable part of the 'coin' in the CoinTetrahedroon a bit smaller.
Decouple (to a large degree) the OpenGL Library from Android.
fix detection of not used attributes.
Major change in JSOn file format:
replace the old (corners,centers) system with a list of VertexEffects which are applied to each cubit variant.
Finally fix the reason foor 'incorrect vertex attribute' message spamming the Log of the Emulator.
Minor.
Debugging the Node piraminds - print the whole piramind along with all the Effects attached to each Node.
Improve the Sink effect so that at extreme strengths (close to 0 and infinity) it inflates and deflates more.Reason: previous version failed to fully inflate the Masterball to a nice sphere.
Remove the requirement that not-postprocessed children in the render scene must be in the first bucket.Seriously simplify renderChildren().
Minor
license
fix for a crasher.
A few code improvements.
FactorySticker: further improvements.
Properly detect which drivers are slow with compilation of TF programs.
Fix for an invalid enum.
New debug to print a whole component of a mesh.
Create the bandaged 3x3 icon: something works, the shape is drawn, but it's entirely black. Investigate.
Further progress with Bandaged 3x3.This requires a new API in DistortedNode to change the Mesh there.
Very important bugfix in the Library: MatrixEffectQuaternion cannot share the tmp static buffers!
new API to enable/disable face culling
Preparation for drawing only the FRONT side of the cubit walls.
Extra API for QuatHelper.
Switch off the OpenGL error debug.
Important bugfix for the 'cube disappears when its mesh gets changed from nice to simple'.
The reason why recreating an InternalBuffer from within the 'create()' function doesn't work but doing the same from 'createImmediately' does work is not fully understood. Nevertheless the fix is correct in the sense that it...
Correct the glBufferData's usage argument: in UBO and VBO cases, it should be GL_STATIC_DRAW.
Add DistortedProgram.stopUsingProgram()
Add a glGetError() call to the main render loop.
Add info about FPS.
Progress replacing the enum ObjetType with the class RubikObjectList.
Adjust ShapeChanging so that it can handle concave cubit faces.Now it is working also in case of the Ivy corner cubits.
Float vertices - scratchbook
Bugfix.
Add to MeshPolygon a special modee to create a 2-band polygon with dense corners (needed to reduce number of vertices in side faces of the TwisttyPuzzle cubits)
Add testing of MeshPolygon to the MeshFile app.
Major change: separate the notion of a TwistyObject and its Node. Now,
1) the Node stays when we change objects (this makes transitions faster)2) it's possible to just create an Object without adding it to the Screen (now app needs to explicitly create the Node and add it to its Screen itself)
Correction
By default, do use Halo Depth (this broke the halo around a solved object in distorted-cube!)
Set possibility for the Postprocessing effects to eclipse the whole object or only work at the edges.
New API: set components centers of all components all in one go.
comment.
New BORDER effect.
Improvements for the new Border effect. Now works, but covers the object - to be corrected.
New postprocess effect 'Border' (not working yet) and adding it to the testing app 'Triblur'.
Mesh creation: finished.
Move destroying most of statics from onDestroy() earlier, to onPause().
Reason: doing this in onDestroy() is too late, because when we have two activities, and switch between them, first the ending's Activity onPause() is called, then the starting Activity's onResume, then it's onSurfaceCreated(), then we insert all the effects, and only then the ending Activity onDestroy() gets fired off - and this might destroy the already enabled effects, like for example the Programs of the Postprocessing effects.
Move QuatHelpeer to distorted-library
Add debugging option to DistortedEffects.
On second thought, come back to Vector. The problem: before we had the potentially blocking function 'effectFinished' inside our synchronized block.
Improve EffectMessageSender: ArrayList with explicit locking in place of a Vector.
1) add missing call to BlockController.onPause() / onResume() to the Tutorial activity2) rearrange BlockController's error reporting to make the case where the MessageSender thread has died separate3) in such case, resurrect the thread4) simplify and harden the EffectMessageSender
More debugs to the EffectMessageSender.
Firebase shows that we have a problem when an app is rapidly restarted, i.e. when is it paused and resumes within tens of milliseconds. Then sometimes the EffectMessageSender thread dies.
Attempt to limit this by synchronizing the whole 'start' and 'stop' methods, although I suspect this will still happen because the very methods are probably in such case called in the wrong order.
Some more debugs for the case when the UI is blocked.