Video yükleniyor...
Video Yüklenemedi
Today we introduce TIN: a powerful and reliable full-text search extension for Postgres. TIN works with complicated WHERE clauses, replication, backups, and maintains correct transaction visibility. It's also mind-blowingly fast.
500,125 görüntüleme • 8 gün önce •via X (Twitter)
34 Yorum

Read more about TIN's performance and how to get started:

🚨 BALLS ON PROD 🚨

You need PlanetScale Postgres to use TIN currently. There’s no public TIN source package or self-hosted installation method documented

I understood Neki not being opensource. For this it's harder to swallow. Will it be in the future ? How can I work with this in a local pg for dev or testing if I can't access the extension ?

Balls

How does it compare with

ah huge. It sucked not having BM25.

sick

@hollylawly can I have a ball chart against algolia please

i wasn't understanding the latency difference until I saw the balls, congrats guys

@chrismunns @paradedb open source 💯

committed rows searchable right away, love that

@samlambert So much good stuff from y’all recently. First Neki now TIN, what’s next?

@paradedb waiting for you folks on this 👀

Oh no not turbopuffer! You were supposed to do Clickhouse

it is fast, 10x is crazy.

@hiradp vector search when

Sure be cool if you open sourced this.

How does it compare to ClickHouse FTS?

having all that work properly with postgres makes it worth checking out

Awesome stuff. Will this extension available outside PlaneScale?

oh yeah, was waiting for this 🚀

Open source when?

Does it work on @supabase ?

Is it bm25?

how does it compares to ElasticSearch?

Great work.

Benchmark metrics look cool, but how does TIN perform when users make frequent typos in the search query? Can it fully replace a pg_trgm + GIN setup for messy search inputs?

hello paradedb prepare yourself and your enterprise license

@BenjDicken please

Fast full-text search is compelling, but the transaction-visibility detail is what stands out. Correctness under replication and backups is the hard part—great to see it treated as a first-class constraint.

How does it compare to `pg_search`

@0xblacklight Really nice write up! Love the details on how to lean into postgres internals to make it fast

the fact that TIN maintains correct transaction visibility is really important for complex queries
