10:05 <doublec> Sad to hear about Causality. So hard to be in the programming language business.

10:07 <doublec> https://www.linkedin.com/pulse/end-causality-constantine-goulimis

13:28 <SeanTAllen> That's very old news.

16:33 <SeanTAllen> Here is my statement on the causality news: https://pony.groups.io/g/user/message/618?p=Created,0,,20,2,0,6182872&offset=0

19:45 <Goba> i think pony should do the same memory model with lifetimes like Rust

19:47 <Praetonus> Goba: could you explain what this model is, for those unfamiliar with Rust?

19:49 <Goba> Praetonus, you have to basically describe with annotations on variables how long they are allowed to live and who gets to access them where. This results in no need for a GC in the language because you are doing what you have to do in C with malloc/free only in a safe fashion because the compiler will check whether the lifetimes you wrote down for all the variables dont intersect

20:15 <Praetonus> I don't really see what it would add to Pony. Reference capabilities already describe how an object can be accessed. For the no GC part, I don't think Pony could work without a GC since actors can send data to each other asynchronously and defining lifetimes at compile time seems impossible. But I don't think that's a bad thing, the Pony GC is really good (based on the ORCA protocol, there is a paper on it). There is no

20:15 <Praetonus> stop-the-world phase and all the reference counting work is deferred. Basically, user code is never interrupted to do GC stuff

21:13 <doublec> I think it's good that Pony sticks with a GC model. It's a nice to have a GC-able language with low level control as an alternative to the non-GC Rust/ATS models.