メンバーシップ
Advanced guides with production-ready code
Thank you for visiting Rork Lab. We publish advanced guides daily with copy-paste ready code, benchmarks, and production design patterns — practical content we hope you'll find useful as the next step after our free articles.
- ✦Unlimited access to all premium articles
- ✦New advanced guides with code published daily
- ✦Production-ready design patterns & best practices
Premium Articles
499 exclusive articlesAfter the Google Play deadline: which dormant Android apps to raise, and which to leave at 35
Apps you have stopped updating are governed by a different bar than apps you still ship. Here is how to inventory effective targetSdk across your builds and decide which apps to raise, hold, or retire.
After Bumping targetSdk to 36, Which Part of Your Dependencies Should You Actually Read
I opened six real dependency AARs and counted what they contribute to my manifest. Not one declares targetSdkVersion, and twelve permissions arrive that I never wrote. Here is what to check before you submit.
The Remote Config keys you add after your release path closes are invisible to the build already on devices
A record of recounting what I can still change remotely before my release path closes. A key you add in the console does nothing unless a build that reads it is already on devices. Which switches to ship in the last build, and a shared-defaults bug I measured.
Two weeks of half-finished batch runs, and every log line said success
A job scheduled twice a day was firing once, and nothing in the logs showed it. Here is the run ledger that catches executions that never happened, plus a backlog-slope check, with real command output.
What Android 16 ignores on tablets, and why app.json gives you no way out
Targeting API level 36 means screenOrientation is ignored on displays at least 600dp wide. Here is how to check your project without a tablet, why app.json cannot express a fix, and what the opt-out property actually restores.
I Split Rork and Claude Code by Who Owns the Build Environment
How to decide between Rork and a terminal coding agent using data from your own repository instead of impressions of generated code. Includes a script that counts which layer your maintenance work lands in, and a check that catches native settings silently disappearing on regeneration.
I Moved the Thumbnails Assuming the Shipped Build Would Keep Asking for the Old Path
Reorganizing image directories on the server does not reach the builds already installed on people's phones. Here is the two-way fallback that returns 200 for both the old and new paths, the ordering rule for images versus catalog, and the four checks I run through a CDN before calling a release live.
Predictive back on Android 16: what to fix before August 31, and what can wait
Targeting API 36 turns predictive back on by default and stops onBackPressed from firing. Here is how I split the work that the deadline actually requires from the work that does not, and how I kept the opt-out from disappearing on the next prebuild.
Every bulk replace exited zero. The damage was in the lines I did not delete
Run a bulk replace over generated code and the breakage lands on the neighbouring lines, not the matched ones. Here is what broke in a live project, and a dependency-free guard that checks the invariants a replace must preserve.
The four acceptance checks I still run after the build turns green
A successful build does not mean a shippable build. Here are the four failure classes an AI build loop cannot see, and a dependency-free script that inspects the artifact itself before you submit.
August 31 Means Something Different Depending on Whether You Plan to Ship an Update
Two separate conditions land on Google Play's August 31 date: API 36 to submit an update, and API 35 to stay discoverable to new users. Here is how I sorted six apps, and who actually gets the extension form.
Three conditions that make Play Policy Insights report nothing on an Expo project
Google Play now ships an open-source policy auditing skill. Running it against an Expo-shaped project, one directory argument moved the result from five detected data categories to zero. Here is when the scan actually reaches your code, and where its output should not be trusted.
The three places I had to fix before a Rork project actually targeted API level 36
My app.json said targetSdkVersion 36. The value my build actually read was 35. Here is the script that reports the effective value, and how I split my apps between raising, leaving alone, and requesting an extension.
When an Expo UI drop-in swap actually removes a dependency
Expo UI went stable in SDK 56 with drop-in replacements for eight community packages. Swapping one import does not always shrink your dependency list. Here is how to decide which swaps actually pay off, straight from the dependency graph.
Find the Code That Trips App Store 2.5.2 Before You Submit
If a remote value can change what your app does, Guideline 2.5.2 may apply. Here is where the line actually sits, plus a script that surfaces the risky paths in your repo before submission.
TestFlight External Testing or Phased Release: Where a Solo Developer's Time Actually Pays Off
When you ship a Rork-built app to other people, you can recruit external testers or let a phased release absorb the risk. Here is how I split my limited time between the two, based on where distribution actually stalls and which three areas still deserve pre-release checks.
Your lockfile's dev/prod split won't tell you which licenses your app must credit
A record of classifying every dependency in a production project to decide what belongs on an app's license screen, and where the copyleft findings and the actual shipped artifact turned out to disagree.
Four Countries, Not the World: Checking Which of My Six Apps the September 30 Deadline Actually Touches
Android developer verification goes live on September 30, 2026 in Brazil, Indonesia, Singapore, and Thailand. Here's how to decide from your own install data whether that date is urgent for you, and what actually blocks indie developers.
Switching to Signed URLs Killed My Image Cache — Decoupling expo-image Keys from the URL
Signed URLs rewrite their query string on every expiry, so a URL-keyed cache never hits. Here is how to pin the key with source.cacheKey, what the cache read and write APIs actually return, and what changed when I re-ran the measurements.
Measuring Hermes Bytecode Delta Updates: One Added Module Cost 29x More Than One Changed Line
I measured OTA delta sizes against a 2.4MB Hermes bytecode bundle. Changing one line cost 2,104 bytes; adding one module cost 61,197. The widely repeated advice about stable module IDs turned out to matter least.
The 429s Kept Coming After Recovery: Measuring Retry Jitter Across 2,000 Clients
My backend came back but the 429s did not stop. I rebuilt the scenario in a 2,000-client discrete simulator and measured what plain exponential backoff, delay caps, 10% jitter, and honoring Retry-After cost in seconds — then reran everything with server capacity that degrades under load.
Cutting MCP Tools Didn't Make Anything Lighter — 2,377 Bytes of Definitions vs 110,298 Bytes of Response
I wrote a minimal MCP server for a wallpaper catalog and measured the byte cost of tool definitions against the byte cost of responses. Here is which side actually matters, and the real reason to merge tools.
Deciding overlay text legibility at ingest time instead of on device — four metrics measured side by side
Moving the question of whether text stays readable over a wallpaper out of the device and into the content pipeline. Four candidate metrics measured across 240 images, including what downscaled judging actually computes.
A JSON File 19% Smaller That Ships Only 3% Smaller — Measuring Catalog Payloads Against gzip
Five payload formats measured on a 20,000-item wallpaper catalog. Key shortening saves just 3.2% after gzip, while a columnar layout cuts 26.8%. Real numbers for Brotli, paging, and JSON.parse time.
Exactly 20 Characters, Still Rejected: Putting Character Counting Behind One Boundary
Emoji and combining marks make your client and your server disagree about length. I measured four counting methods on Node v22 and found 75.6% of naive truncations split a grapheme and 36.1% produced replacement characters over UTF-8. Here is the shared-module design that fixed it.
When 1.10.0 Gets Locked Out: Measuring Four Version Comparison Approaches for Forced Updates
A remote-config gate for minimum supported version, rebuilt after measuring four version comparison approaches. Includes the cases where localeCompare reports equality, the fail-open boundary, and the incident caused by gating on a build still in review.
When Two-Character Queries Silently Return Nothing: Measuring Japanese Search Indexes in an Expo App
SQLite FTS5's trigram tokenizer returns zero rows for Japanese queries shorter than three characters, without raising anything. I benchmarked linear scan, a bigram inverted index, and FTS5 over a 20,000-item catalog to find the real threshold.
Every Experiment Kept Landing on the Same Devices: Hash Choice and Salt Position, Measured Across a Million IDs
On-device experiment assignment looked fine in isolation, then collapsed the moment two experiments ran side by side. Here is what one million IDs revealed about four hash functions, why the culprit was the concatenation order rather than hash quality, and the implementation I settled on, with the measured numbers.
My generated wallpaper catalog hashed differently on every run — isolating five sources of nondeterminism
The same image folder refused to produce the same catalog.json twice. Directory enumeration order, JSON.stringify key reordering, collation, Unicode normalization, and async completion order — measured one at a time across 1,225 files until the output settled on a single hash.
What Renovate may bump in an Expo app, and what it must never touch
Turning on automated dependency updates in a Rork-generated app also hands Renovate the 123 packages Expo SDK 57 pins. Measured on 2026-07-30, six of them sit a full major version behind npm latest. Here is how to generate the ignore list from the SDK instead of maintaining it by hand.