Pinned Tweet
We've built Postgres driver for libraries 馃
We didn't want to, but we had to:
鈻笌 all major runtimes support(rip deno)
鈻笌 tls/http/ws/dataapi transports
鈻笌 data correctness(oh man...)
鈻笌 native Effect
鈻笌 explicit batch, pipeline, parallel apis
鈻笌 bulkInsert, bulkUpdate, copyFrom
鈻笌 metrics/telemetry
鈻笌 for libraries - Drizzle, Kysely, Prisma, etc.
鈻笌 effeciency and performance
鈻笌 replication, backfills, snapshots
let's dive in 馃У
This and procedures could be the biggest dx and perf gainers in your project
Can we at least all agree shipping abstraction on top of SQL is substantially harder than a sims-like game with frontier realism?
Ok, really curious to try Bun AOT now
We shall compare it to rust
Replying to @DrizzleORM
New benchmarks 馃殌
we're now smoking GoLang, thanks to Bun and Node ofc
we're just minimising overhead
We run benchmarks on the same i3 Lenovo since Aug 2024
we went from 7.9k rps on Node 22 + Drizzle v0.33 to 23.9k rps on Bun 1.4.4 and drizzle today
We've been running our driver on internal projects for over 4 months so far, it powers drizzle studio, drizzle pulse and drizzle cdc
Give it a try with [email protected] and please share some feedback!
We are not here to rewrite everything in Rust, we're here to improve existing solutions, make them correct and fast!
Thank you 鉂わ笍
We will be dropping Drizzle Cache API in v1
Caching query responses in ORM is:
鈻笌 wrong place
鈻笌 easiest footgun
鈻笌 nightmare for us, library maintainers
let me know if you use drizzle cache API, we will propose you a proper way









