hey all :wave: I'm currently familiarising myself ...

# arrow

[Marcus Ilgner]

hey all 👋 I'm currently familiarising myself with Arrow and now I'm wondering how one does control concurrency when using its high-level concurrency. It's great not having to care about these details when implementing business logic but at some point it would be nice to control for example the size of the thread pool. Can you recommend any resources that go into more depth on that topic?

[CLOVIS]

The functions you write are suspend . suspend functions can only be called in suspend functions. Of course, we need to be able to call the first suspend function, and we do this on a CoroutineScope, which contains a CoroutineContext . The thread pool is configured as part of the Dispatcher , one of the configuration elements of CoroutineContext .

[CLOVIS]

More info → https://kotlinlang.org/docs/coroutine-context-and-dispatchers.html

[Sam]

It’s important (ish) to distinguish between concurrency and parallelism. Suspend functions can add concurrency by splitting a computation into multiple sub-computations. Arrow’s parMap function has an optional parameter to limit the number of sub-computations to create, which might be something like what you’re looking for. But parallelism is not introduced until you actually run the function, and the level of parallelism is controlled by the CoroutineDispatcher.

[Sam]

Practically speaking the level of concurrency is an upper limit on the level of parallelism — i.e., if you only break the task into ten pieces, you can’t usefully increase the parallelism higher than ten simultaneous threads

[Marcus Ilgner]

good point 👍 indeed, I was thinking more about parallelism there. Just want to build a clearer picture in my head of when and how code will actually be running in the system when put together using these mechanisms. But these points already clarified a great part of that. Thank you both!