2
0

Polish.txt 2.6 KB

1234567891011121314151617181920212223242526272829303132333435363738
  1. --------------------
  2. Delay this until release:
  3. Input issues:
  4. - I cannot have input of 1 one frame and input of 0 another. This happens when frames are too short. I might need to add very short smoothing times?
  5. - I cannot have input set to 1 during the entire frame if the frame is very long and user just moved the mouse a bit.
  6. - This might be even more of an issue with joystick axes as I currently have no mechanism of knowing what % of the frame are they being moved.
  7. - Maybe: Ability to define input sampling rate. Input sampling then runs on its own thread at 120/240/etc/Hz. Every frame the collected values are averaged.
  8. - This could be a part of OIS. Mouse sampling should already be doing something similar.
  9. - I can't think of ANY other way gamepad axes can work with long frame times, other than this approach
  10. - Check DirectInput::Poll
  11. - And DirectInput::GetDeviceState to get immediate mouse data
  12. - Check if this state is an absolute value, difference from last call or what.
  13. - Potentially also GetRawInputData as I might be able to use that instead of windows messages
  14. - Probably remove UE4 smoothing as I'm not even sure what it does
  15. Problem with current OIS mouse approach is that if I call capture() too often, no new sampling data is available
  16. but the mouse might still be moving and it just reports 0. Which is wrong if the mouse is moving.
  17. - These is no way around this if querying the state too quickly. I can either keep the current system and remember
  18. last 5-10ms of input and average that, or I can use the above solution with polling which should have the similar effect.
  19. -----------------
  20. Not so critical
  21. - I have a feeling DX9 might start crashing once I resize to sub-fullscreen res and go back to FS - test it
  22. - Add VSync toggle to RenderWindow if it doesn't already exist.
  23. - Test creating a fullscreen window on window creation.
  24. - Refactor GUIElement & GUIElementBase
  25. - Sprite rendering methods should probably be marked as internal (possibly others too)
  26. - A lot could be made private
  27. - Has two methods for retrieveing ElementType
  28. - getVisibleBounds doesn't make sense
  29. - There's still a crash regarding an uninitialized mCachedPtr on a C# class when shutting down. Attempt to find consistent repro steps.
  30. - Perform optimizations as needed to get reasonable performance
  31. - Consider removing SceneManager from Core? I have two scene manager base classes.
  32. - GpuParamBlockBuffer and GpuParams::cloneForCore are very clumsy in how they deal in keeping two separate param blocks. Rethink this a bit.
  33. - Test texture compression and mip generation with DX9 and OpenGL