Post #1839170

2026-04-07 17:04 UTC

@matthew @rhysmorgan @akbashev no I think this is a super-good question! In this case, we're talking about a decision that has pretty extensive technical implications. And real, large code bases tend to benefit even more from relaxing constraints. So, I would argue in favor of some upfront learning in exchange for significant longer-term wins.

Replies (2)

@rhysmorgan@mastodon.social 2026-04-07 17:07

@mattiem @matthew @akbashev OK, I'm glad you've posted this, because my thinking is on the right track! Definitely going to need some time for re-learning things, but that was going to be necessary as Approachable Concurrency inevitably becomes default anyway. My idea is that bigger code bases with more tests will benefit even more from this approach, because you have fewer constraints on where you can use types, less MainActor contention in tests, and it'll be easier to tell…

Open ##1839171
@matthew@mastodon.me.uk 2026-04-07 17:51

@mattiem Relaxing constrains is probably the best option. Sadly, I don’t see an overwhelming appetite for learning, as many people are happy to ship code if the pipeline is green. (And if the tests don’t pass, re-run them until they do pass) There’s little thought about code architecture, and certainly not threading/isolation.

Open ##1839176