Daniel Hall

@DonnyWals And yet, amazingly, it’s still a total ripoff for developers that offers no benefit other than the “privilege” of being one free app out of millions. Or the honor of being a paid app that gives 30% of revenue to Apple while being otherwise ignored. All for a bargain $99/year!

Did you know that Apple's developer program hasn't seen a price hike in more than 16 years? It used to cost at least $499 to develop for the Mac, and $99 to develop for iPhone. In 2010 Apple reduced the cost to $99 for both. Today, the cost is still $99. Taking inflation into account that means the cost (in the US) has roughly been halved in the past 16 years.

@bxlewi1 I completely agree. Hard to express a strong argument against the flaming pants assertion.

@_danielhall Onboarding on every platform has way too many vague spots or things that simply don’t work.

Apple software and usability is total absolute flaming pants.

@zats @krzyzanowskim I really enjoy SwiftUI overall, but your point is 100% valid. I challenge anyone to consistently predict what a chain of modifiers will / won’t do without previewing or building to see what happens at runtime (i.e. “The JavaScript Experience™️”)

@krzyzanowskim Whole architecture of modifier is baffling to me, I am sure you can arrive at the current solution if design from scratch, but it feels so bizarre For all the crap iOS devs give JS, we ended up with a UI framework where you can call modifier on any view and it *might* work

SwiftUI.Text is an exciting type in the SwiftUI framework. text modifiers are implemented on a Text, and the same modifiers are implemented on View. One returns Text type, and the other returns "some View". It has it's confequences, there's a reason, also there is a duplication. the consequence that hurts me: valid: text.italic() + text.bold() impossible: text.modifier(Italic()) + text.modifier(Bold()) because ".modifier())" does not return Text 🥁 and that is what I need! so I have to reimplement half of the SwiftUI now I guess.

Engineers who are incapable of trusting others inevitably become bottlenecks. Therefore orgs who keep engineers with trust issues or a low opinion of others in key positions will always move slowly.

@birbirisidir @crypto_bonesNFT @KevinNaughtonJr If the the business needs changed, it’s not a refactor, it’s a new feature or a rewrite. The problem is when unit tests have to be modified as part of a refactor. Then you can’t tell if the existing behavior was truly maintained or if there has been a regression.

@birbirisidir @crypto_bonesNFT @KevinNaughtonJr Not exactly — it should cause your existing test to fail, not require the test to be modified. If the existing test can’t validate the refactored code, how do you know that the code meets the existing requirements? Only the code should be fixed until the existing test passes.

unit tests are a waste of time

@birbirisidir @crypto_bonesNFT @KevinNaughtonJr Changing a test is the same thing as writing a new test. If a unit test can’t remain constant when the code it tests changes, then it’s not a refactor it’s a rewrite. That means more risk as well as a major purpose of unit tests being lost: the ability to refactor safely.

@crypto_bonesNFT @KevinNaughtonJr Then, just change your unit test?

I still don’t understand—what is the benefit of having AI help you write something that has already been written? Are our jobs simply to generate pages of safely plausible text, not to create and describe new ideas and solutions?

@BigCatJared Haha — that would be truly game changing!

@_danielhall Maybe it will write perfect code that does not need to have that context later. 🤔🤣🤣🤣🤣🤣

What people get wrong about ChatGPT writing code is that they only show examples of generating new code for a given prompt. That’s cool but 95% of coding is writing within the context of existing code. AI isn’t super useful until it can write correct code for any given codebase.

When this happens, the wrong thing to do is to double down on solving your internally-created problems in ways that will become new problems. Instead, realize that you are focused on the wrong problems entirely & radically reassess how to streamline and get back to the right ones

At Twitter, I would estimate that at least 80% of iOS dev time and effort were absorbed by internally created processes, tools and problems which had little or nothing to do with creating good user experiences. 4/🧵

An informative article about the evolution of the Facebook iOS app & architecture. While the process of evolution is unavoidable, I’m struck by how each solution becomes the next worst problem & how much work went into fixing the wrong problems 1/🧵 engineering.fb.com/2023/02/06/ios…

@DXZDB @elonmusk I mean, yeah it was totally a joke based on $TWTTR hitting one of Musk’s haha price points back then. Not at all a serious prediction, but I’m still willing to retroactively claim credit for calling it! 😉

Whoa! Was I the first person to predict @elonmusk buying Twitter?!!

@bob_burrough Interesting perspective! Wouldn’t that at least replace human jobs in the Growth, BI, Revenue sorts of areas? Or do you just see AI as always being limited to a tool used by a human, rather than a replacement for a human function?

@_danielhall We will definitely get to a world where corporate AI will compete against other corporate AI to maximize profits. IMO, that will not threaten humanity in any way. It should be no different than the introduction of any other technology into business.

For decades, we were deeply skeptical whether computers could play chess competitively. The world was blown away when, in 1996, a computer beat the chess chess world champion. Today, computers are consulted by top grand masters to find novel attacks against their opponents.

@bob_burrough That’s because it’s a hobby though, isn’t it? And not a survival necessity? If a company commercially produced winning chess games, it would almost certainly use AI players and not humans in order to have the most successful business model.

We should also acknowledge there are still chess grand masters. They didn't become obsolete. In fact, chess is more popular now than at any time in history. Chess AI isn't an existential threat to human players. On the contrary, it helped make chess accessible to more people.