Interesting article from JVM world - Conservative GC: Is It Really That Bad?

[Dmitry Olshansky]

On Monday, 5 March 2018 at 02:42:48 UTC, Nikolay wrote:
> These guys have independent JVM implementation and used 
> Conservative GC for many years. As I can see it it is very 
> similar to d-runtime GC. But their conclusion is: "sooner or 
> later the absence of knowledge about the liveness of local 
> variables will lead to problems in production that just cannot 
> be handled"

Provided that you manage resources via finalizer and there is no 
RAII and the like.

Interestingly enough both new low-latency GCs for Java are 
non-generational.

>
> BTW Have we any progress or scheduled for GC improvements (e.g. 
> process fork for parallel GC, thread-local GC)?