Questions on multithreaded design

[yohannd1]

Hello! I'm experimenting with designing a terminal text editor in Janet. This is my second attempt, with the previous one being stalled due to issues I had with... I think blocking reads?

Because of that, I'm experimenting with making a multithreaded design, with one thread for the editor logic, and another for I/O with the terminal (rendering UI and reading keyboard input).

• What would be the best way to share a text buffer between the threads? I've seen something can be done with ffi/malloc but I haven't thought of a way to lock the buffer so the memory is thread safe.

• Is there a standard way to communicate between threads? Something like an RPC-ish API. That's what I'm thinking of implementing using threaded channels, at least.

[psagers]

• What would be the best way to share a text buffer between the threads? I've seen something can be done with `ffi/malloc` but I haven't thought of a way to lock the buffer so the memory is thread safe. • Is there a standard way to communicate between threads? Something like an RPC-ish API. That's what I'm thinking of implementing using threaded channels, at least.

I think you'll find the answer to both questions in the ev module. Specifically: https://janet-lang.org/api/ev.html#ev/lock https://janet-lang.org/api/ev.html#ev/rwlock https://janet-lang.org/api/ev.html#ev/thread-chan Note that using locks to coordinate threads is a time-honored tradition in systems programming, but also surprisingly difficult to get fully correct in nontrivial situations. Hence the emergence of higher-level abstractions such as channels and actors.

[yohannd1]

Ah, I hadn't seen these were available, thanks! And I will take a look at actors, hadn't heard of these before.