Exactly, Sparkling uses reference counting, and one of the major TODOs is the introduction of weak references in order to address the issue of cycles.
Implementing weak references sounds like a feature that will make the language's source code much more complex than the simplicity emphasized in your goals.
nim has some sort of automatic cycle detection. You might want to look into how they do it. It doesn't hurt to check it out.
Excellent! I was hoping that the answer would be that there is no GC.
How is reference counting deterministic in a multi-threaded environment?
> Another thing is garbage collection. It seems to be the silver bullet of automatic memory management when it comes to scripting languages, but it has some definitive, serious downsides. Non-deterministic behavior is one, speed is two, memory overhead is three, the difficulty of a decent implementation is four. Reference counting is a simpler approach that is more lightweight, easier to implement, completely deterministic and it has no memory allocation overhead (a program uses only as much memory as needed). Refcounting has its disadvantages too (e. g. memory leaks caused by cyclic references), but I believe that its good properties outweigh them.