igalklebanov

As far as query building goes, `kysely` is more in-depth, type-safe, expressive, 1:1 and doesn't require importing as many things.

`kysely-ctl` doesn't generate migrations for you, and `kysely` not being an ORM, doesn't provide declarative schema design.

Hey

DISCLAIMER: I'm co-leading the org/project.

We recently broke 6M downloads per week on NPM, and became 3rd after `drizzle-orm` and `@prisma/client`.

If you haven't tried it yet, it's a query builder, not an ORM. You don't outsource your SQL to someone else. It's type-safe, like.. it's super important to us. You can use it with ORMs - e.g. Prisma, mikro-orm, zenstack, etc. Allows you to compose some complex stuff but keep it maintainable af.

If you have. Great seeing ya'll here.

0.29 was a real nice release, with lots of goodies. Can't wait for 0.30, gonna be super fun.

can you share your personal documentation story? what did you not find? what was painful?

thoughts? requests? issues?

<3

JOOQ is cool!

What do you mean by "better type safety"?

Kysely's ecosystem has 2 codegen libraries:

kysely-codegen introspects the database directly.

prisma-kysely generates types based on Prisma schemas.

Looks cool!

Kysely was inspired by Knex, but took a more predictable route instead of providing generic APIs that produce different SQL for different dialects. We believe that dialect "lock-in" should never be a real concern because migrating from let's say, MySQL to PostgreSQL is extremely rare in the real world.

<3 from Kysely.

Would love to have a discussion about your use case in our discord. We're constantly thinking about improving that part of our API as it's at the heart of many things.

<3 from Kysely. kysely-codegen is dope!

I'd suggest you try them both, and pick what you like better or what feels safer to bet on for your project.

I know some of our users use both, zapatos for codegen and kysely for querying.