Gwen Shapira

Poster sessions are honestly my favorite for complex deepdive topics. It is visual, conversational, collaborative and includes coffee. Just how I like my deep technical content.

PGConf.dev 2026: top things to do at the conference:
• Debate MVCC trivia with strangers
• Discover a feature you didn’t know you needed
• Realize you’re not the only one who reads pg_waldump for fun

Registration and CFP are open.

CFP for PGConf.dev 2026 is open until Jan 16 🐘

Tips for strong proposals:
• Share what you discovered. Not just what you built.
• Be specific
• Discuss bigger Postgres picture
• Failures welcome

Be specific, connect to Postgres internals, and don’t hide the messy bits.

My favorite ritual on the calendar :)

Heh! Love it 😍

At what point do we stop saying “worth its weight in gold” and start saying “worth its weight in high-end GPU”?

So.. Pierre Barre just built an NFS server implementation on SlateDB. ZeroFS is today days old, but it's going to get wild. Bottomless storage and checkpoints mean instant FS snapshots (zero overhead).

github.com/Barre/zerofs

Love it! Such a gorgeous collection.

It’s been 22 years and I’m still madly in love with my husband.

I use a spreadsheet to track all my documentation

It's important to keep all your docs in a row

Maybe it’s from reading the news? Social media?

Definitely. Descript is such a time saver for these things.

YouTube video by PostgreSQL Development Conference Re-engineering Postgres for Millions of Tenants (PGConf.dev 2025)

Videos from PgConfDev are starting to show up on YouTube!

Great time to catch up on both the new content and the amazing talks from past events.

You can start with my talk: "Scaling Postgres to Million Tenants"

www.youtube.com/watch?v=EfAS...

Never mind. Looks like I remembered linear algebra wrong and matrices are strictly 2-dim.

Why aren't these called Matrixes? Like multi-dimensional vectors were called back when I did linear algebra?

So, Tensor (as in Pytorch) is just like a vector, except that it is somehow better for GPUs? Did I get that right?

I made a note to try this next time I have a problem query to debug :)

Nice! I wasn't familiar with this paper. Thank you!

🛠️ Unused index debug flow:
ANALYZE ➜ rethink predicate ➜ simplify query ➜ audit policies & GUCs ➜
If still wrong, try pg_hint_plan (force an index to prove a point) or hypopg (hypothetical indexes) before filing a bug.

1️⃣ Stale stats: ANALYZE hasn’t run; filter looks un-selective.
2️⃣ Truly un-selective: Stats are right; scan really is cheaper.
3️⃣ Complex estimates: Many joins / ORs / CTE / functions scramble row estimates.
4️⃣ Handcuffs on: RLS limits optimizations or GUCs forbid the path.
5️⃣ Planner bugs: Rare but...

“Why is Postgres seq-scanning rather than use this index?” 🤯

Short answer: the planner thinks that index won’t help (or can’t use it).

Longer answer—5 usual suspects 👇

omg, this is so purrfect!

You are not wrong 😻

Cats are great, actually.

New blog: Using Postgres pg_test_fsync tool for testing low latency writes

An easy to use tool that comes with standard Postgres installation, for testing write latency of devices where you put your WAL/redo logs and any other files requiring low write latency.

(link in reply)

I feel like I've found a new controlling meme for open source contribution and innovation.

Thank you! I'll check out context7

Context7 - Up-to-date documentation for LLMs and AI code editors Generate context with up-to-date documentation for LLMs and AI code editors

I've registered it with context7 so it's now available as up-to-date nile documentation via the context7 MCP context7.com/www.thenile....