On Monday, 6 June 2016 at 15:06:49 UTC, Carl Vogel wrote:
(...) Also, the idea that more people will adopt D if you just "get rid of the GC" ignores the fact that you don't just "get rid of the GC," you replace it with another memory management scheme (manual, RAII, RC). If you replace all the parts of the language and phobos that rely on GC with, e.g., a poor implementation of RC, then you're just going to get another round of complaints, and no real adoption.
I think you are going to get some adoption it you replace good GC with clunky RC. The key difference is a call to a function that uses GC is incomplete: some of it will execute later, after the call has finished. On the other hand a call to an equivalent function that uses RC has only local (in time) consequences: once you finished the call, it stopped executing. If it returned something that needs to be freed later, you know what it is. Of course people can write arbitrarily messed up things like singletons etc. but I'm not counting those because good libraries are usually free of those. This means you have control over all the OTHER code, however inefficient the call is. Hence, the second is acceptable in low-latency code, but the first is not.
