After my previous post I got nerdsniped by@alexmillerdb.bsky.social about what an optimal IR for dbs would look like and since i am on vacation it's starting to look like a very long read

finally managed to finish a draft i started working on a month ago about Inkfuse, continuations and query execution paradigms deferworks.org/posts/inkfus...

Nice example of evaluating filters on compressed data: This post from Martin Prammer and @spiraldb.com is a spiraldb.com/blog/pattern...

Pattern Matching on Compressed Strings without Decompression

Evaluating SQL LIKE directly on FSST-compressed strings: a symbol-aware automaton and a SIMD prefilter let Vortex match without decompressing the column first.

spiraldb.com

Make TPC-H data scale factor 10 faster than you can type the command (under 10 sec). The new release of tpchgen-cli is out -- try it via uvx uvx tpchgen-cli parquet --scale-factor=10 Huge thanks to @clflushopt.bsky.social and @kevinjqliu.bsky.social

Mike Matheson just rips one through traffic and it's 2-0 #Habs

With the next product launch season fast approaching, a reminder of the first question you should *always* be asking when a figure of merit is presented without context: #HPC

Big Data on the Move: Can a Framework Laptop 13 ultrabook run terabyte-sized workloads with DuckDB? @szarnyasg.org ran the experiments and shared his finding in our latest blog post: duckdb.org/2025/09/08/d...

As someone who has exclusively used Rust, Go and C++ for the past 8+ years I find the JVM stack (lots of Java & Scala) a breadth of fresh air from a tooling and productivity perspective.

I think I have finally managed to articulate this very fundamental but slightly counterintuitive fact about how shell redirection works

Stopped once I hit the 18 seconds total runtime, code is available here github.com/clflushopt/s... might take this for another tour, still lots of tricks to pull (custom hashtable for the joins, morsel driven parallelism...)

@clflushopt@clflushopt.bsky.social · last yr.

My work on SIGMOD 2025 (unregistered) continues my implementation now runs in 24 seconds instead of 11 minutes on my hardware (Ryzen 9 9950x), next goal is sub 10s. I will write a blog post about this, since I was led down some dangerous paths hint blog.llvm.org/2011/05/what... #SIGMOD2025

My work on SIGMOD 2025 (unregistered) continues my implementation now runs in 24 seconds instead of 11 minutes on my hardware (Ryzen 9 9950x), next goal is sub 10s. I will write a blog post about this, since I was led down some dangerous paths hint blog.llvm.org/2011/05/what... #SIGMOD2025

I am not having fun trying to reverse engineer and implement the TUM's db group paper on hash tables for join processing db.in.tum.de/~birler/pape...

@apachedatafusion.bsky.social supports a fully extensible range of user defined window functions as well as scalar, aggregate and table functions. Aditya Singh Rathore's new blog has more: datafusion.apache.org/blog/2025/04...

A great question -- and one that I used as a foundation for an entire talk speakerdeck.com/bcantrill/th...

Things I Learned the Hard Way

Talk given to the engineering team at Rokt on November 9th, 2023. Video: https://www.youtube.com/watch?v=bJ0y7Oqr4Zo

speakerdeck.com

Dave Farley@davefarley77.bsky.social · last yr.

What's one thing you know now that you wish you learned earlier in your software engineering career?

can I code fast? no. but can I code well? also no. but does my code work? alas, no

hello