@dillon_mulroy Automating that "Engineer 2" code review + automatic merge risk assessment to only require human review on more risky PRs should help with this. At least that's the direction we're taking on my team. The current state is of humans proxying AI output is ridiculous.
@dcatalin98@mattpocockuk Planning requirements was always "a" hard part, but its also incomplete. For a sufficiently complex problem, I dont think anyone can accurately predict every details, edge cases and decisions. A lot of those are discovered during the implementation.
@matthewcp Looks good! I'm glad to see robot making a comeback after all these years, definitely gonna try and use it. Btw the links under "Core concepts" don't seem to work, at least on mobile.
@DavidKPiano For sure! Whenever I get confused about what I'm building, I break it down into its different states and events. Even if I don't end up using a state machine, this helps me reason about my code tremendously.
@Rich_Harris I've definitely seeked complexity for its own sake when starting out, mostly to understand things and learn as much as possible. And it did feel nice to be able to handle such complexity. But the more I gain experience, the less I need to do that.
@thomasglopes@radix_ui@SvelteSociety The article is great, learned some nice stuff. But it seems to me that "as" is dismissed too early. I believe it's possible for the "as" prop to either be a string or a component, and if it is a component, the props and events contract can be enforced with TS, at least in Svelte.