Asynchrony Is Not Concurrency

10 months ago

The blog post distinguishes between asynchrony, concurrency, and parallelism, providing clear definitions for each.
Asynchrony is defined as the possibility for tasks to run out of order and still be correct, while concurrency is the ability of a system to progress multiple tasks at a time, and parallelism is the ability to execute more than one task simultaneously at the physical level.
The post highlights the confusion in common programming lingo where 'concurrency' is often used to describe both asynchrony and concurrency, leading to a lack of understanding and suboptimal software ecosystems.
Zig's approach to async I/O is highlighted, where asynchrony does not imply concurrency, allowing synchronous and asynchronous code to coexist without requiring code duplication or forcing users into async-only models.
The post introduces `io.async` and `io.asyncConcurrent` in Zig, explaining how they handle asynchrony and concurrency differently, with `io.asyncConcurrent` ensuring tasks run concurrently when required.
The author emphasizes the benefits of separating asynchrony from concurrency, such as avoiding library duplication, improving user experience, and preventing deadlocks.
A live demonstration is promised to showcase Zig's async I/O redesign, including implementations of thread pools, green threads, and non-concurrent Io.

Hasty Briefsbeta