Jonathan Joelson

@PavelASamsonov@mastodon.social I’ve found that, in the long run, fast, cheap, and good are actually all correlated.

@schwa@mastodon.social @alexr@mastodon.online It had the same issue as many car touchscreens: it replaced proven useful physical controls instead of augmenting them with contextual touch controls.

@mattiem @Cykelero I wonder if there a term for when an underlying system gets so complex that the nice clean abstraction on top of it ceases to be useful. I feel like we’re at that point with both memory usage and drive space.

@nicklockwood@mastodon.social Something that came up at work just today: apparently TextField *still* doesn't have a non hacky way to set a maximum number of characters. Yeesh.

@nicklockwood@mastodon.social The degree to which SwiftUI still falls on its face with more complex controls after all these years is disturbing. Makes me think there are some fundamental limitations to the whole architecture. TextField is another one that is not fit for purpose as soon as you move beyond the most basic usage.

A weird thing about Swift Observation is that there's no way to express in a property's type signature whether it tracks updates or not. Anyone else think that's kind of a weird choice on Apple's part? #Swift

@marcoarment @lukaskubanek This is wild. Don’t important Apple apps like Notes and Reminders use CloudKit?

@peter@thepit.social I’m sorry that you’re encountering this. imo it’s a terrible sickness in the job market that someone starting their career applying for a junior role would need to “demonstrate expertise”.

I know people have always uncritically copied code from StackOverflow, but at least there you have competing answers, voting, reputation, comments, a date on the answer so you know if it's potentially out of date, etc. LLMs for answering programming questions feel like a big step backward to me.

Just encountered someone uncritically posting a screenshot of an #LLM answer to explain a #SwiftUI bug we were seeing, and it suggested a significantly disruptive workaround.

Turns out its diagnosis was completely wrong, and actual human SwiftUI experts found a very simple solution to the problem in 5 minutes.

I'm sure there are lots of people out there leveraging LLMs responsibly, but imo this is a seriously dangerous tool for non-experts who can't critically evaluate the answers.

@gruber @ismh The analogy is a bit flawed in that drivers and cyclists do have to share the same limited public street space, and so there actually is some genuine conflict and tension in designing streets for cars vs bikes.