Boom ๐ฅ
Paperback book just went live on Amazon. Paperback took a lot of work so I really hope you enjoy it :)
Paperback: https://t.co/Jjqt8IrITJ
Ebook: https://t.co/kA000X5O66
#iosdev#swiftui
@abdy424@neerajjj6785 I don't have any talks but I usually write most of the code by hand and then ask AI to review it. Then I read the review and see if I can improve my code.
It really depends. I think in most scenarios, if you are fetching extra fields even text based it would not matter that much. On-device SQLite is blazing fast so you won't feel the performance hit.
However if you are pulling a lot of extra fields and then not really displaying or using them in maybe in those cases you can create a relationship.
In SwiftData you can use propertiesToFetch to retrieve only the properties you need instead of loading all the properties.
One minor issue is that... It does not work :)
It still fetches all the properties.
That make sense. Bad architecture is going to pay the price in time. Tightly coupled systems will take longer to update or modify.
I guess we can even say that now these principles, patterns and clean code is even more important because we can tell our agent to follow good practices and we can't tell them to follow good practices if we don't know about them ourselves.
When developers write code manually, they are more involved in making decisions, exploring different approaches, and understanding trade-offs.
This process helps them understand the problem better.
When AI agents write the code, we risk missing out on that learning experience, even if the generated code works perfectly.
@uliang6482 Walking away and thinking about the problem is still part of critical thinking, which is what you are giving away with AI since AI is thinking for you.
I agree with clients pushing for unrealistic deadlines. We all have been in those situations.
You can def do all those after the code is generated by AI but I bet most developers don't do that. Once they see all green, they move on.
A hybrid approach can be to write code manually... keep thinking about alternate solutions. Once the solution is achieved review it with AI.