@lassiecoder Gradle errors are where a lot of RN builds go to die, so this fills a real gap. Does it read EAS Build logs too? Downloading the log from a failed cloud build and getting the actual cause would save Expo folks a lot of scrolling.
@jeff_okoli Love the blur. Gotcha for anyone copying it: on iOS listen for AppState 'inactive', not 'background', or the switcher snapshot beats your blur. Android never fires 'inactive', so a JS overlay often loses that race. FLAG_SECURE (expo-screen-capture) is the reliable fix there.
@nativecookdev Would love to see real numbers on this. @oceandawy any chance you measured cold start (tap to first screen) and memory after a few minutes of use, RN vs Exact, on the same mid-range Android and an iPhone?
Underrated file-based routing win:
New devs stop asking “where do I add this screen?”
Create the file at the path. Done.
Most of the “you registered this in the wrong navigator” PR comments disappear with it.
Your app shouldn't look like every other AI app. 🎨
Meet Theme Editor. Pick a preset or describe a vibe, and every screen of your React Native app restyles at once.
New engineers judge a stack on day one, not month three.
If their first hour with Supabase is spent getting containers running, that’s the impression they keep, even though the problem was never Supabase.
Make the first query happen in the first five minutes.
Most “EAS build failed” threads quote the last 20 lines of the log.
The real error is usually higher up, in the first phase marked red (often “Install dependencies” or “Run gradlew”).
Expand that phase and search for “error”, not “failed”. Saves a lot of guessing.
Writing one Edge Function shouldn’t need a cloud account, a CLI login, a secrets store, and a dozen containers just to run it locally.
Deploy to the cloud. Develop on your laptop. One process, real Postgres, functions included.
Same patch, two channels:
OTA update: 70% of users had it in 24 hours.
App Store release: 30% in a week.
The rollout channel is a product decision now, not just an engineering one.
Asked 6 developers why they skip local Postgres and build straight against hosted Supabase.
Every answer was about Docker on their Mac, not Postgres.
That's a packaging problem, not a database problem.
Saw an AI-built app rejected four times under Guideline 2.1 (App Completeness).
The agent built every screen with data in it. The reviewer installed fresh, had no data, and saw blank screens.
The edge-case screens are the review.
@NesanSelvan04@supabase The 'local dev without Docker' line is the biggest unlock here the 20-container compose stack was the top complaint I kept hearing. Declarative schemas 2.0 plus OrioleDB in public beta is a serious iteration on the DX. Curious whether this changes the self-hosted story too
Watched checkout user tests this week.
The biggest conversion lever wasn't the payment gateway. It was how fast the payment sheet appeared after tapping Pay.
A web-view spinner is where people quit. Measure time from tap to sheet, not just server response.