News for hackers. Stuff that matters.
Re:Java 12 by pron on Tuesday March 19, 2019 @10:13PM (#19438270)
What makes you think reference counting leads to shorter pauses than tracing? While it is absolutely true that sophisticated implementations of garbage collection via ref counting can exhibit similar performance to sophisticated tracing algorithms[1], most implementations of ref counting aren't nearly as sophisticated as modern implementations of tracing, and tend to be worse.

[1]: https://www.cs.virginia.edu/~cs415/reading/bacon-garbage.pdf

Re:Java 12 by tambourine_man on Wednesday March 20, 2019 @12:26AM (#19438986)
Not necessarily shorter, but predictable pauses.
Re:Java 12 by pjmlp on Wednesday March 20, 2019 @03:59AM (#19439912)
Very predicatable indeed.

Go and C# running circles around Swift.

"35C3 - Safe and Secure Drivers in High-Level Languages"

https://www.youtube.com/watch?v=aSuRyLBrXgI

Possible stack overflows and unexpected pauses due to heavily nested data structures.

"CppCon 2016: Herb Sutter “Leak-Freedom in C++... By Default."

https://www.youtube.com/watch?v=JfmTagWcqoE

Re:Java 12 by pron on Wednesday March 20, 2019 @07:18AM (#19440936)
If your memory usage is regular and predictable, then both ref counting and tracing would behave regularly and predictably, and if it isn't, neither is.