Reference counting in general has been an interest of mine since even when I'm writing server-side software it's usually pathological-case latency rather than raw throughput that really matters.
It's a pity that it's mostly C++ (smart pointers) and Obj-C (ARC) that have access to this.
( I say this realizing Python is hybridized. )
Edit: Most of my hesitations regarding using the JVM have to do with how incredibly finicky tuning it can be.
Reference counting comes with a huge toll on interoperability. Forget one Py_INCREF(Py_None) in a tight loop, and you're up to a segfault in a place very, very far away.
It's a pity that it's mostly C++ (smart pointers) and Obj-C (ARC) that have access to this.
( I say this realizing Python is hybridized. )
Edit: Most of my hesitations regarding using the JVM have to do with how incredibly finicky tuning it can be.