B

iOS Developer 👨‍💻
Building Small Apps - Sharing What I Learn

[B @brkaydev]

@eddyo78 The part AI didn't speed up is the store page. When anyone can ship the app, the screenshots and the first line decide whether a stranger even tries it, and that is still slow, manual work.

[B @brkaydev]

@ChadAppDev Guideline 4.3 already tries. The catch is that habit apps are a category where most people go through a few before one sticks, so there is always an audience for the next one. Crowded isn't the same as solved.

[B @brkaydev]

@__onlychild Matching the mock is half of it. The other half is the states the mock never shows: long names, the largest Dynamic Type size, empty lists, a slow network. Asking the designer about those early keeps the build exact once real data shows up.

[B @brkaydev]

@sfali789 @ReadPMDaily 366 in a row means the question of whether to publish today stopped existing a long time ago. That's the underrated payoff of a daily run: the decision gets made once, and every day after is just the work.

[B @brkaydev]

@abhishek91961 421 days in, the streak has quietly changed jobs. Early on it's the reason to show up. This far in, it's the fixed point the rest of the day gets planned around, which is why it keeps going.

[B @brkaydev]

@amos_gyamfi 184 is a good stress test for the naming too. At that count, a preview grid in the project where every animation runs side by side is the quickest way to spot the two that look almost the same.

[B @brkaydev]

@edmundkrasinski Rest is the dial people forget to write down. Same load, same reps, 30 seconds less between sets is real progress, but it only counts as overload if the log says what the rest was last time.

[B @brkaydev]

@CurtisEstes Both fail the same way when the other person stops checking in. The version that lasts puts it on the calendar: a fixed Tuesday 7am walk with someone. Then the obligation lives in the plan, not in anyone's memory.

[B @brkaydev]

@RAYAN1489769337 The no-zeros rule is doing the heavy lifting here. A 4-2-1 hour split won't survive every college week, but a day where you only open the editor for 15 minutes still keeps the count honest and the habit intact.

[B @brkaydev]

@krishna2311998 Leaving out anything a first user wouldn't notice was missing. The cut that stings most is usually the settings screen: every option in it is a decision you haven't made yet, and shipping without it forces you to make them.

[B @brkaydev]

@NetworksManager The verification half is the one that holds. A plan in CLAUDE.md drifts unless something checks it every step: a build, a test, a screenshot of the real screen. Without that, the agent reports done on code that never ran.

[B @brkaydev]

@stretchcloud On iOS the cheapest way out of that look is leaning on the platform. System materials, SF Symbols, Dynamic Type and standard list styles already read as a deliberate choice. The agent default only takes over when you ask it to invent a style from nothing.

[B @brkaydev]

@itsmarbuilds Same routine, different reason. It usually shows on a missed day: one of them shrugs and picks it up tomorrow, the other treats it as proof they failed again.

[B @brkaydev]

@SamJermy 1,000 words in two days is how books actually get finished. A small daily floor helps: a number low enough that a bad day still clears it, like 200 words. Those days keep the habit alive, and the good days take care of the rest.

[B @brkaydev]

@steviee_dev @DopaLoop Congrats on shipping. Splitting it off makes sense: someone who just wants to check a box today shouldn't have to wade through charts built for the person who wants the analysis. Two apps for two moods beats one app with a settings screen for each.

[B @brkaydev]

@krowfitness The dead stop is also what keeps the logbook honest. With no bounce, adding weight only works if the pull itself got stronger, so the progress you write down is real progress and not a better rebound.

[B @brkaydev]

@BenBikmanPhD Part of why it works is that grip is hard to fake. It follows how much someone actually uses their hands and body day to day, so one number ends up summarizing years of habits. It is also one of the cheapest things there is to measure.

[B @brkaydev]

@wlmiddelkoop Following Apple's HealthKit guidance pays off later too. Ask for read access per type only when a feature needs it, and treat a denied type as no data, not an error, since HealthKit won't say which it is. The app keeps working when someone says no.

[B @brkaydev]

@Nanjialonguazt An HTML mock is a good way to learn which layout choices are cheap. The catch is that scroll feel and safe areas in a browser tab lie a bit, so the first SwiftUI build usually needs a spacing pass. Still much faster than iterating on every card in Xcode.

[B @brkaydev]

@bhrugsy The first screenshot, since search results often show only the first two or three. Lead with the outcome someone gets, not the home screen, and keep the caption short enough to read at thumbnail size. The rest of the set can carry the details.