john-chapman
commented
6 years ago The implementation would simply convert quat -> mat3 and then call the existing Gizmo functions, so explicit support for quaternions isn't strictly necessary. Indeed, overloading the Gizmo functions is difficult since all the arguments are float*.
Converting quat -> mat3 and back would obviously incur some precision loss however it's probably acceptable (and the return conversion can be avoided if the Gizmo function returns false).
http://www.euclideanspace.com/maths/algebra/realNormedAlgebra/quaternions/functions/