🚀 go-pugleaf

Randell Jesup writes:

> But GC is unlikely to reclaim that storage.  Yes, the graph no longer
> has pointers to the disconnected subgraph.  However, the subgraph may be
> circular (you said it was cyclic), so all the nodes of the subgraph may still
> have references to them.  Can current GC systems recognize such a loop and
> remove it in it's entirety?

Of course.  Only brain-dead garbage collectors that rely exclusively on
reference counts can't get rid of cycles.  (Reference counting garbage
collectors are bad news anyhow, since they entail the overhead of a ref
count inc/dec on every assignment.)

> How complex can such a loop be and still be recognized?

Arbitrarily complex.

-- Steve