Jelte Fennema-Nio

Not about tech this time: I recently donated some of my stem cells to a patient with blood cancer, who I don't know, somewhere in the world. You too can register yourself to save a life: NL=matchis.nl/word-stamceldo…, USA=nmdp.org/get-involved/j…, and Google for other countries

@mim_djo Only slightly too soon. The next release will have PowerBI support (probably deployed today/tomorrow). It's already working in our staging environment.

@JelteF spoke too soon, getting full wire compatibility is a pain

This has been my last few months: Getting MotherDuck to speak the Postgres protocol! There's still a long road ahead to support all the BI and schema management tools. But I won't rest until \d works! But first some parental leave. motherduck.com/blog/motherduc…

I'm now officially a Major Contributor of Postgres: postgresql.org/community/cont…

I did a talk about the main project I've been working on for the past year: "pg_duckdb". So if you're curious what that is, give it a watch: youtu.be/0kc-W5o0tcg #posetteconf

🎸 Tune in to (@JelteF) Jelte Fennema-Nio's talk during #PosetteConf Livestream 2️⃣ to learn about the new pg_duckdb open-source #Postgres extension that makes it easy to run analytics queries straight from within Postgres.🦆 posetteconf.com #postgresql #duckdb

Just finished recording my talk for POSETTE about pg_duckdb. You can watch the live stream together with me on June 11th. Because it's recorded I'll be able to constantly answer questions during the stream. More info: posetteconf.com/2025/talks/pg_…

@gwenshap You could even use a DOMAIN type to wrap the ENUM to make sure the CHECK constraint which handles removals holds in all your tables. Reordering of existing values can be done using remove+add. Which is a hassle, but reordering is also difficult with your other suggestions.

@gwenshap Not saying enum is without issues, but renaming values is definitely possible with ALTER TYPE name RENAME VALUE... Removing is indeed not possible without recreating it, but you could rename a value to "__removed1" and still use a check constraint to ensure that it's never used.

🚨 Postgres' ENUM type is a trap. It looks appealing if you use enums in your code, but... ❌ You can’t remove, rename or reorder values. Only way is to drop and recreate the type... which you can’t do if any table is using it. 🔒 Adding a new value takes a lock. 🤯 Tight DB ↔ app coupling. Add a new enum value? Your app’s type is now out of sync. You must deploy schema and code at the exact same time. Terrible for independent deploys. And remember - you can't roll back this change, because you can't remove values from the enum. ✅ Better: use a CHECK constraint or a small reference table. You get validation and flexibility without all these gotchas.

Today I finally tried out the new "Instant SQL" feature that @hamiltonulmer has been teasing for months. The immediate feedback really feels like MAGIC. For more fancy GIFs check out todays release blogpost: motherduck.com/blog/introduci… Be sure to try it out, you won't wanna go back

PgBouncer 1.24.1 is out. It fixes CVE-2025-2291, which allowed expired passwords to still be accepted. It also fixes a few issues introduced in 1.24.0. github.com/pgbouncer/pgbo…

@mitsuhiko Too tired from every year a new python installers toolchain. So now I'm sticking with plain pip an requirements.txt

If you use python, but you don't use uv yet: why?

@emrahdx Many people that have their transactional data in PG don't want to set up replication and a whole separate DB only for doing analytics. So they run analytics in Postgres. That's slow, but if you don't have a ton of data it might be fine. pg_duckdb allows PG to go even further.

@JelteF I still don't get the why.. 🤔

pg_duckdb 0.3 was just released! Major features include: Support for using PG indexes, a much easier syntax for querying parquet/csv/iceberg files in blob storage, and last but not least: ✨A shiny new logo✨ Changelog at: github.com/duckdb/pg_duck…

@onyekaugo And we have this blog, but that's still using the old read_parquet syntax. I'll try to get that one updated soon. motherduck.com/blog/pgduckdb-…

@onyekaugo We have these docs in case you haden't found them yet: github.com/duckdb/pg_duck…

I had a great discussion with @thegeeknarrator about putting DuckDB into Postgres using pg_duckdb. You can watch it here: youtu.be/Qn9_J_HI53o

@CarLaTeXSE That project is pretty inactive. It's not completely unmaintained though. If something is broken on e.g. a new Python version I'll fix it. But there's no new features planned. It's fairly complete though in my opinion, so you can still use it.

@JelteF Hi @JelteF, I'd like to know if PyLaTeX is still maintained or is an inactive project. I'm writing here because I don't know how to contact you. Thank you.

PgBouncer 1.24.0 is out, with many improvements. And prepared statement support is now turned on by default! github.com/pgbouncer/pgbo…

It was lots of fun talking about pg_duckdb on @PostgresFM

pg_duckdb v0.2.0 is out: Now with Delta Lake, JSON, Azure and transaction support github.com/duckdb/pg_duck…