sutton037

People often assert that scope-based reference counting such as shared_ptr in C++ collects garbage “promptly” and some people define this as “collects at the earliest possible point”. For example, at the time of writing the Wikipedia page about garbage collection says: “ Compared to tracing garbage collection, reference counting guarantees that objects are destroyed as soon as they become unreachable … ”  – Wikipedia Similar claims can even be seen in published research such as the paper “ Down for the Count? Getting Reference Counting Back in the Ring ”: “Of the two fundamental algorithms on which the garbage collection literature is built, reference counting has lived in the shadow of tracing. It has a niche among language developers for whom either performance or completeness is not essential, and is unused by mature high performance systems, despite a number of intrinsic advantages such as promptness of recovery and dependence on local rather than global state.” – Blackburn...

Although the vast majority of programmers have now migrated to garbage collected languages and will probably never go back, there are still a few clinging to manual memory management. In most cases, the continued use of manual memory management is for good reason but some of these people are perpetuating myths in an attempt to justify avoiding garbage collection. Determinism can be a genuinely good reason to stick with manual memory management and is practically important in memory-constrained embedded devices. However, C++ programs are not as deterministic as people sometimes claim and, in particular, thread-safe reference counting using  shared_ptr  is non-deterministic. Specifically, threads holding references to shared reference-counted objects race to perform the final decrement and the thread that wins the race is responsible for destruction.

In a recently-posted video , Herb Sutter (a prominent C++ expert) describes his favorite C++ 10-liner as “a thread-safe reference-counted object cache”: shared_ptr<widget> get_widget(int id) {   static map<int, weak_ptr<widget>> cache;   static mutex m;   lock_guard<mutex> hold(m);   auto sp = cache[id].lock();   if (!sp) cache[id] = sp = load_widget(id);   return sp; } This example is very interesting. Firstly, it manages to pull in reference counting, weak references and a mutex which are all very rare in modern programming. Secondly, it contains a memory leak that is difficult to fix in C++ because APIs are burdened with memory management details and this API is incapable of expressing deterministic cleanup because there is no facility for a widget's destructor to remove its entry in the map. Finally, the correct name for this data structure is a concurrent weak dictionary , specifically one with weak values. You'll find corre...