🇺🇦 Oleksii T.
banner
oleksify.me
🇺🇦 Oleksii T.
@oleksify.me
Programmer, Designer, Musician. Vegan for the animals. 🤘🐷🐮🐔🐹🐱🐶
Very odd article. Or rather some personal thoughts. So much hate, but really no value. Still, fly.io can’t make a stable hosting platform even after so many years. With that attitude - I now understand why.
June 5, 2025 at 6:29 AM
We tried electrisql and it’s a pain to implement besides their toy app examples. Write examples even for toy apps are complex enough. And you still need client to be JS - not LV.

In this regard so far only zero.rocicorp.dev worked out of the box as expected. It does mean going away from LV though
zero.rocicorp.dev
January 18, 2025 at 6:41 AM
We solved first one partially with Web Components, but the second one is looking nearly impossible.

4/4
January 17, 2025 at 10:33 PM
- Offline Mode - this is a big problem. I don't see any solution so far, that's why we thinking researching writing whole frontend client using Sveltekit + some Local-First stuff (like Zero Cache) for data replication.

3/4
January 17, 2025 at 10:33 PM
- Server has no idea about the client even though it is doing work for the client. This means DOM environment - viewport stuff, portals, all the APIs you need to build complex UI components.

2/4
January 17, 2025 at 10:33 PM
We're writing complex ERP with LiveView and we've been through all stages of hell initially, but it's settling down now. We started really eraly adopting LV, and now ended up building all components as Web Components with LV wrappers and hooks.

But the 2 main pain points are always the same:

1/4
January 17, 2025 at 10:33 PM
Somehow it does not make it better in any way...
January 16, 2025 at 1:48 PM
We switched to Neon now, and really happy about it. It’s cheaper to Supabase (supabase PITR costs $100 alone) and feels faster. Also autoscaling is really nice and reliable.
December 15, 2024 at 1:40 PM
To be fair, their support is good and helpful. Everything basically is great, except reliability.
December 15, 2024 at 12:48 PM
Yeah, it was their postgresql service. They warn it’s not managed db, but still. It doesn’t mean it can just disconnect. While I waiting for support answer I set up new cluster and migrated db, because prod can’t wait. We never figured what happened. Also, it was just one of many incidents.
December 15, 2024 at 12:41 PM
Once prod db just got disconnected, and we could not connect it back. I see the pg service in admin, but fly machines don’t. After that I said enough is enough. I love the system they built - it’s amazing. It just unreliable as hell.
December 15, 2024 at 12:33 PM
I have the same thoughts. We where using their services for about a year, and problems where happening all the time. Sometime db just got disconnected or disappeared… Switched to Render an no issues now.
December 15, 2024 at 11:44 AM