Vishram used to take ~3s to show its first screen. Now it's ~0.6s.
I didn't remove anything. I asked one question of every startup task: can the first screen exist without this?
If yes, it moved off the launch path.
#Flutter#buildinpublic
The 4 things that did it:
1. Critical path only. Only what the first frame renders blocks launch
2. Cache + skeletons. Show saved data instantly, sync after
3. Parallel + fire-and-forget. Independent tasks run together
4. Lazy loading. A section loads its data when you open it
My handle is in the Super Productivity v19.0.1 release notes.
The feature: double-click projects, tags, and folders in the sidebar to rename them. Small change, but everyone who renames anything touches it. ๐งต
It's my 7th merged PR there. Others: i18n for hardcoded strings, a screen-reader label, an invisible placeholder, a misleading (0) on empty subtasks. Also named in August's Contributors of the Month.
Unsexy PRs compound. #opensource
If you're in India and Vishram was stuck loading or showing "tap to retry" โ that's what this thread was about. It's fixed now.
https://t.co/pAzGutTYv9
If it still misbehaves on your network, I want to know.
1/ ๐งต
It worked perfectly on my phone.
That's the whole reason this bug took months.
Some users had movie screens hang for 2 minutes, then fail. Others never saw it once. Same build, same screens.
Four rounds of fixes later, the bug was never in my code.
12/
The Worker sits on @Cloudflare's edge, reachable on the ISPs that block TMDB directly. Forwards every request unchanged.
Bonus: it injects the API key server-side, so the key doesn't ship in the app anymore.
After four rounds of fixes, the change was a base URL swap.
11/
Which left the answer I'd been avoiding.
I didn't want a proxy. I'm a solo dev โ one more thing to deploy, keep alive, monitor, and debug at 2am when it turns out to be the broken part.
But the domain was blocked. That's it.
App โ Cloudflare Worker โ TMDB