From Side Project
to Google Play
Building the app turned out to be the easy part. Snekcade, a retro 8-bit game that teaches Python, was playable for weeks before it was launchable. This is the part nobody puts in the tutorial: payments, email, the domain, the Android wrapper, and the long list of declarations Google wants before anyone can install your app.
What Snekcade Is
Snekcade turns a Python course into eight pixel worlds and 42 short levels. Each level is a few swipes: read a tiny idea, predict what the code prints, put lines in order, then write real Python that runs on the phone. There is an arcade mode, daily review, a monthly league, trophies, and a final world where you build the logic of your own Snake game and then play it. The first worlds are free with no account. A subscription, the Snek Pass, unlocks the rest.
Technically it is deliberately small: a static progressive web app with no framework, a handful of serverless functions on Vercel, and Supabase for accounts and sync. Python runs in the browser through WebAssembly, so there is no code execution server to secure or pay for.
Payments: Plan B Is Part of the Plan
Selling software internationally from Serbia means you want a merchant of record: a company that sells on your behalf and handles VAT, invoices and chargebacks. My first choice reviewed the store and rejected it, with no reason given. The integration was finished and tested by then, which hurt.
I switched to Polar, which supports sellers in Serbia. Because the app only ever stored a normalized subscription row (status, renewal date, end date), swapping providers meant writing one new webhook handler and one checkout endpoint, not touching the game. A few rules kept that layer honest:
Email Without Paying for a Mailbox
An app with accounts needs two kinds of email: a public address people can write to, and reliable outgoing mail for sign-in codes. The registrar wanted a yearly fee for a mailbox I would barely use, and its DNS editor was locked behind a paid tier. So the domain's nameservers moved to Vercel, and the mail was split in two:
Android Without Rewriting Anything
The Play Store version is a Trusted Web Activity: a thin Android shell that opens the web app full screen in Chrome, plus the Play Billing bridge so subscriptions bought on Android go through Google. One codebase, two stores.
The shell only goes full screen if the website vouches for the app through a Digital Asset Links file containing the signing certificate fingerprint. The catch: with Play App Signing, Google re-signs the app with its own key. The fingerprint of the key you upload with is not the one users get, so the file needs both. Miss the second one and every user sees a browser address bar at the top of your "native" app.
The first uploads also bounced on requirements that change every year: the minimum SDK level for the browser helper library, the minimum Play Billing Library version, and the target SDK level for new apps. None of it was hard, but each one cost a build.
The Play Console Gauntlet
A new personal developer account starts with identity verification, a verified phone, and proof that you own a real Android device. Then the app itself needs eleven declarations before the first review. Most are a few clicks; a few deserve thought.
Where It Stands
The web version is live, payments work end to end, and the Android closed test is in review with Google. Next is collecting testers, running the fourteen days, and applying for production. The lesson I keep relearning: the code is maybe half the work of a product. The other half is trust plumbing, the payments, email, signatures and declarations that let strangers rely on it.
If you have an Android phone and want to help, play the web version or send me the Gmail address you use on Google Play at hiss@snekcade.com and I'll add you to the test.