Pinned Tweet
🧵I ran a test to find out how Vercel Fluid compute behaves and here's what you must know:
1/5
When using Fluid Compute (I think enabled by default now), it seems like one could reuse a Singleton, like e.g. a db connection with postgres.
This is true AND false. 👇
DAVID retweeted
Today, we're introducing Vector Buckets.
A new storage option that gives you the durability and cost efficiency of Amazon S3 with built-in similarity search
supabase.com/blog/vector-buc…
My life broke apart. My 12 years relationship is no more.
The last 4 weeks felt like 2 years of growth and pain. And I'm slowly getting back to work, realizing that if I let myself down now, I will loose what's precious to me: my work.
Life goes on but it's okay to feel
DAVID retweeted
we have released a stripe sync-engine, which synchronizes your Stripe account to a Postgres database (in real time)
run it as a full server or install it as a typescript library running in a serverless function
"I'm scared, didn't get salary for 2 months, 700 bucks left" - a good friend of mine.
This is one of the reason why I became self-employed - I can be fully transparent with myself
"Hey there, just wanted to check and see if you're interested in boosting your social media presence with our AI Tool, 14 days free of charge!?"
What vibe coders can't comprehend: Actually inventing something
JSON is token‑expensive for LLMs – just like @mattpocockuk frequently mentions.
Meet TOON, the Token‑Oriented Object Notation.
💸 40–60% fewer tokens than JSON
📐 readable & tokenizer-aware
Wrap your JSON with `encode` to save half the token cost:
github.com/johannschopplich/…
Next.js, by now, has one of the biggest configuration jungles. Starting from next.config over to export const dynamic over to use cache
I went with @tan_stack / @reactjs SPA + @honojs by now (Hono has built-in tRPC btw hono.dev/docs/guides/rpc)
That's pretty shitty. But if we consider the fact that supabase/storage has an S3-compatible layer, couldn't we just use that to self-host S3?
Peak sexy: built db-monitoring for my client to send problematic queries to Grafana in Real-time
Now you can just zoom in Grafana on any time range and see related long runners
I don't like bait.
There is no 4$ Hetzner VPS with 4 VCPUs and 8Gigs of RAM - it's 6,59$.
Also, that doesn't include backups, and you're a fool not to activate those, which brings you to 7,90 USD total.
While AWS is down I'm sitting here with my $4 Hetzner VPS, self-hosted landing pages and even self-hosted Supabase pulling 100k+ Steam requests per day without a single interruption.
People have no idea how powerful it is to be in control of your own infrastructure.
I could literally boot up an old PC with one of the daily backups and continue to run uninterrupted even if Hetzner went down.
Might make a guide for this at some point. Everything is way too centralized today.





















