Here are the spirit that Soya follows (or should follow ;-) ):

• Soya's goal #1 is to allow to develop as rapidly as possible 3D games and other 3D apps (Rationale: Soya targets "amateur" developpers, who code during their limited free time! Moreover, rapid development allows to gain a precious time that can be re-invested in improving your code or testing your app. For game, it leads to a better gameplay)

• Soya's goal #2 is to be as easy as possible to learn, in particular for people with no 3D background at all (Rationale: Soya is not a toy; to be easy to learn is only goal #2 since any newbie will, a day, not longer be a newbie)

• Despite its simplicity, Soya never sacrifices performance and speed (Rationale: speed is important for 3D game!)

• Soya's API does not necessarily fit to the mathematical, computational or technical reality of the 3D, e.g. Soya API does not require the use of matrices (Rationale: math are not the natural way to represent 3D object. Soya should be useable without an important mathematical background. However, Soya provides matrices, mainly for debug purpose)

• Soya relies a lot on Python facilities and modules, e.g. saving 3D model is done with the pickle module. As a consequence, Soya won't evolve toward a multi-language 3D engine and will stay Python-centred (Rationale: Python modules ease the Soya development, but also the use of Soya, since they are already well-known to Python developers)

• Soya has plenty of dependencies (Rationale: any good OS has a package system tools today)

• Soya always assumes by default the most common usecase (Rationale: doing so lead to a substencial time gain)

• Soya is an "atypic" 3D engine and relies on a certain number of controversal choices that are somehow debatable ; however, most of these choices are deliberate (Rationale: Soya should be seen as a "research project" aiming at "a new way for 3D" ; the objective is not to satisfy anyone, but to be the ideal 3D engine for a few persons. Python does similarly)