theanirudh

He means that they moved from AWS to Planetscale and yes it handles much larger DBs.

Their databases are hosted on AWS and GCP so latency isn't much of an issue. They also have AWS Private Link and if configured it won't go over the internet.

The issues weren't PlanetScale related. We use Hasura and when we did the cutover, we connected to the DB via PGBouncer and some features don't work right. Started seeing a lot of errors so paged them and they helped out. We were connecting directly to PG previously but when we cutover we missed that.

We just migrated to PlanetScale Postgres Metal over the weekend. We are already seeing major query improvements. The migration was pretty smooth. Post-migration we hit a few issues (turned out it wasn't an issue with PlanetScale), and the PlanetScale team jumped in immediately to help us out, even on a Saturday morning so support's been amazing.

The Insights tab also surfaced missing indexes we added, which sped things up further. Early days, but so far so good.

This used to be the case when the platforms were much younger and new features were introduced every year. But I would say it’s mostly stabilized now. React native had day 1 support for Liquid Glass and new AI APIs introduced on iOS 26.

In English, silence is transcribed to "Please like and subscribe"

Seems like a lot of extra work in cases where we change the scoring mechanism, we will then have to invalidate the existing entries, recalculate and write it out again compared to just having an endpoint that will take all previous lessons and generate the next lessons on demand.

How do sync engines address issues where we need something to be more dynamic? Currently I'm building a language learning app and we need to display your "learning path" - what lessons you have finished and what are your next lessons. The next lessons aren't fixed/same for everyone. It will change depending on how the score of completed lessons. Is any query language dynamic enough to support use cases like this? Or is it expected to recalculate the next lessons whenever the user completes a lesson and write it out to a table which can then be queried easily?

Do they find ARC AGI also tough due to the same reason? I’ve seen some examples where the input was ASCII art versions of the actual image.