Conversation
|
@ArthurRibeirox Thanks again for putting this together — you were absolutely right to call out the use of double-based math in some of the operators. That’s something I want to eliminate to keep everything fully deterministic, so great catch 👍 I’m actually planning to mark the float operators as obsolete and move them into a later breaking release. Before I merge this in, could you retarget the PR from main → dev? Once that’s done, I’ll pull it in and apply a few follow-up tweaks (mainly around consistency across operators and tightening up a couple edge cases). Really appreciate you taking the time to dig into this — solid contribution. |
|
Yes, my first idea was to remove the operators, these conversions should always be explicit. |
2 participants
Hello, I'm still learning about Fixed math, but from what I understand some of the operators of Fixed64 are wrong.
They are converting Fixed64 to double, performing the operation, then back to Fixed64, there are 3 possible non-deterministic operations right?
I changed so every operator converts all to Fixed64 and perform the operations on that class.
Some tests had to be changed to FuzzyEqual, and I'm not particularly happy about the
FixedQuaternion_ToEulerAngles_HandlesGimbalLockPitch, but did not know if there was some explicit handling of that case I was not aware of.Let me know if there is anything else I need to change!