iiinova — where your digital self keeps evolving

Alejandro Wainzinger @xevix.bsky.social · 02/10/2026

Having fun testing Nimble on ollama (open weights jev basically), and works as advertised. Can't wait for an official GUI, but a quick streamlit hack works just fine for now.

000 0

Alejandro Wainzinger @xevix.bsky.social · 16/09/2026

New version of DuckParq now with Vortex support. 1.5B rows / 22.3GB Vortex file opens instantly. Yes, try not to make files that gigantic, but in case you do 😶‍🌫️. github.com/xevix/duckpa...

010 0

Alejandro Wainzinger @xevix.bsky.social · 13/09/2026

Quick hack to pipe results from nushell to a duckdb table for further querying. gist.github.com/xevix/466e5e...

020 0

Alejandro Wainzinger @xevix.bsky.social · 02/09/2026

Zarr dataset (~1B numbers, (24, 37, 721, 1440) float 32) mean() CPU vs GPU. Manual threadpool gets CPU close to GPU but numpy (1 thread) and dask are quite slow. M4 Max 128GB. MLX code requires so few lines, definitely worth it. gist.github.com/xevix/c320ed...

000 0

Alejandro Wainzinger @xevix.bsky.social · 01/09/2026

Love the technical write-up, lot of deep work went into this. Gemini translation for a TL;DR: CTE go brrrr 🚀

020 0

Alejandro Wainzinger @xevix.bsky.social · 30/08/2026

Interestingly even a COUNT(*) that is done entirely from metadata is faster, because Parquet's Thrift metadata deserialization takes variable time, whereas Vortex's FlatBuffer footer is zero-copy, no per-chunk deserialization, constant and takes microseconds.

100 0

Alejandro Wainzinger @xevix.bsky.social · 30/08/2026

vx is a nice first party tool to look around where the FastLanes bitpacking is happening and how.

110 0

Alejandro Wainzinger @xevix.bsky.social · 30/08/2026

Quick test of Vortex vs ParquetV2 w/ zstd on taxi data in DuckDB. 48M rows 1 file. As advertised Vortex has faster selective querying and aggs due in part to avoiding decompression (~30% faster here). File also ~7% smaller. Will test Hive once partitioned writes fix released (fixed upstream).

110 0

Alejandro Wainzinger @xevix.bsky.social · 28/08/2026

Neat talks hosted by Motherduck yesterday. I’ve been interested in zarr and using it in SQL w/o xarray for messing with geolibre things, cool to see. Also good to hear Motherduck taking the AWS news well, hopefully bodes well for the future of DuckDB.

131 0

Alejandro Wainzinger @xevix.bsky.social · 26/08/2026

DuckDB extension to add Hive data summary clause using the PEG tokenizer. Can't quite use the PEG grammar yet I think, but here's what it could look like when released. Going to be cool to see what people come up with when it's this easy to modify the grammar. github.com/xevix/duckdb...

000 0

Alejandro Wainzinger @xevix.bsky.social · 02/08/2026

I wanted a parquet viewer like Tad but it hasn't been updated in a while. Not bad for a few hours vibecode in Opus 5. Swift, powered by DuckDB via C API. Supports modifying underlying SQL, seeing schema.

100 0

Alejandro Wainzinger @xevix.bsky.social · 16/03/2026

DuckDB just casually in the nVidia keynote. AI doesn't make databases irrelevant, it makes them more relevant than ever.

072 0

Alejandro Wainzinger @xevix.bsky.social · 07/02/2026

Claude cowork w/ Opus 4.6 is definitely smart, but got stuck on a data task, I stopped it, pointed it to DuckDB, done instantly. LLMs still have much to learn 🤔

100 0

Alejandro Wainzinger @xevix.bsky.social · 23/01/2026

Great talks at South Bay Systems hosted at databricks on xNVMe, fast SSD query processing, and using NPUs for DB work. Much work using DuckDB extensions. Need for async I/O as bottleneck a common topic, mainly at larger scale. luma.com/8a54z94d?tk=...

010 0

Alejandro Wainzinger @xevix.bsky.social · 04/10/2025

Taking the DuckDb hoodie on a trip. Not exactly Amsterdam but I’ve heard they like columnar databases here too.

030 0

Alejandro Wainzinger @xevix.bsky.social · 25/08/2025

Compiling DuckDB on Windows 11 (ARM) using UTM VM on macOS to debug Windows compile issues. It's a shame msvc doesn't exist outside of Windows, mingw/clang don't work the same and cross-compiling is tricky. Compiling takes 5-10 mins (instead of 1-2 mins native), but it works 🎉!

130 0

Alejandro Wainzinger @xevix.bsky.social · 15/08/2025

Just the 1 day of data above is ~125GiB compressed, ~585GiB uncompressed. One month is about 3.75TiB compressed, or 17.5TiB. It makes sense this dataset is so popular for testing and analysis, wow.

060 0

Alejandro Wainzinger @xevix.bsky.social · 15/08/2025

Stretching DuckDB w/ Common Crawl, ~1.7B rows, ~300 parquet files. ~2-3s for single-column aggregations, ~2-3 mins to SUMMARIZE the data, peaking at ~12-14GB memory usage. Not exactly real-time, but the fact you can do this on a laptop with no server setups or Spark pipelines is still amazing.

1439 0

Alejandro Wainzinger @xevix.bsky.social · 12/08/2025

Neat little hack to get Hive partition list in DuckDB, useful for an overview. Might be neat to have built-in. gist.github.com/xevix/04f33d...

140 0

Alejandro Wainzinger @xevix.bsky.social · 29/06/2025

Automator using simple shell script to call sqlfluff. Added keyboard shortcut for the service too. Easier than making browser extensions for each browser, although unfortunately not cross-platform.

000 0

Alejandro Wainzinger @xevix.bsky.social · 29/06/2025

Added an Automator quick action to run sqlfluff for formatting SQL in browser fields, used here in the DuckDB UI. Only needs sqlfluff, optionally configure rules. Would be cool to get built-in one day, but works for now.

100 0

Alejandro Wainzinger @xevix.bsky.social · 21/06/2025

Vibe coding NOAA GHCN weather visualization from scratch w/ Claude Code and DuckDB MCP. There's Evidence and other vis tools but I don't want a pre-cached set of data, I want it to query live. Cool that this can be put together w/o writing my own HTML/CSS, as a web backend dev 😅

110 0

Alejandro Wainzinger @xevix.bsky.social · 08/06/2025

Veo 3 to generate a duck writing SQL. Bias is still an issue, but pretty amusing.

100 0

Alejandro Wainzinger @xevix.bsky.social · 07/06/2025

Been trying to use Cursor+Gemini Pro 2.5 in agent mode to add a new feature. It's gone into a lot of loops, and then fell into full-on desperation 😂. Laughing pretty hard at its attempts. I think software engineering is safe for a bit longer.

010 0

Alejandro Wainzinger @xevix.bsky.social · 06/06/2025

Tested Claude Desktop DuckDB MCP on 275 years of weather parquet data on my local filesystem and asking for analysis, and it concluded human-caused global warming unprompted 😶‍🌫️. Cool that it could tell when to check filesystem and when to use DuckDB for SQL querying.

000 0

Alejandro Wainzinger @xevix.bsky.social · 28/05/2025

A brief attempt at differentiating between DuckDB scenarios now, as well as Motherduck. This might help in trying to understand potential use cases for each.

010 0

Alejandro Wainzinger @xevix.bsky.social · 17/05/2025

Wanted to catch up on duckdb developments between 1.2 and 1.3 (upcoming), lots of great work it seems. Created a log with just commit messages and asked Claude to summarize. Not too bad, but anyone know a better way? git log --pretty=format:"%s" --since "2025-02-04" > commits.log

010 0

Alejandro Wainzinger @xevix.bsky.social · 09/03/2025

Fantastic weather in Silicon Valley lately. Getting very green.

000 0

Alejandro Wainzinger @xevix.bsky.social · 04/03/2025

Compiling DuckDB on Steamdeck from source took under 11 minutes with ninja. The fans were blaring but CPUs did not melt… I think 🤔 . Will make a short script of dependency installation later.

260 0

Alejandro Wainzinger @xevix.bsky.social · 27/02/2025

DuckDB on a Steam Deck? Why not. Anecdotal testing shows it’s about 3x slower than M1, 6x slower than M4, but still subsecond queries on millions of rows. The new shell install script worked like a charm too 🎉

2233 0

Alejandro Wainzinger @xevix.bsky.social · 16/02/2025

Generic LLMs seem to have trouble coming up with GROUPING SETs as an optimization to some aggregations, and have a strong bias for UNION ALLs and complicated CASE statements. I wonder if top of the line paid LLMs fare better.

000 0

Alejandro Wainzinger @xevix.bsky.social · 15/02/2025

Needs more training data on other dialects of SQL. Seems to default to PostgreSQL a lot, not up to date syntax/features of other DBs, even before its cutoff date.

100 0

Alejandro Wainzinger @xevix.bsky.social · 15/02/2025

It nails understanding text to SQL (~70% on Birdbench) and it shows, although sometimes needs to simplify queries a bit. Explanations are ace. Follow-up questions dig into detail. Definitely the way to go to write complex analytical queries before manually doing so.

100 0

Alejandro Wainzinger @xevix.bsky.social · 15/02/2025

I've been continually impressed with qwen-2.5 coder months later at figuring out differences in queries, and solving some not-impossible but non-trivial queries.

100 0

Alejandro Wainzinger @xevix.bsky.social · 12/02/2025

Conversion to a single Parquet V2 w/ zstd yields 162.4Mb compared to 65.8Mb grib2, which shows the efficiency of grib2 for raster data (matrix data, instead of columnar). Ordering by Hilbert curves led to worse compression in this case. Still, not too bad for 100M rows.

100 0

Alejandro Wainzinger @xevix.bsky.social · 12/02/2025

Plotting 1M points of grib2 data by first using duckdb to convert grib2 to parquet w/ geom column using spatial extension, geopandas in Jupyter. Raster to vector not super efficient, but at least possible. Amazing in just a few lines of code. Data belongs to ECMWF. gist.github.com/xevix/83ad1e...

110 0

Alejandro Wainzinger @xevix.bsky.social · 12/02/2025

Indeed, looks like DuckDB has taken the lead back since this post by DataFusion. Great to see ongoing improvements all across the data ecosystem. datafusion.apache.org/blog/2024/11...

011 0

Alejandro Wainzinger @xevix.bsky.social · 09/02/2025

Testing spatial extension in DuckDB, here's creating a triangle from lon/lat points and outputting GeoJSON. GDAL is used under the hood and I could use it directly, but something magical about not having to write appcode for this. Will test GeoParquet processing in the future.

040 0

Alejandro Wainzinger @xevix.bsky.social · 06/02/2025

I asked Google AI Studio to summarize the commit messages listed in the DuckDB 1.2 GitHub release post. Insane amount of feature work and improvements. github.com/duckdb/duckd...

110 0

Alejandro Wainzinger @xevix.bsky.social · 30/01/2025

Love the upcoming feature showing human-readable large numbers in DuckDB. format_bytes() is great for byte sizes, and there's ways with functions to work for numbers, but having it done automatically is a fantastic QOL change.

040 0

Alejandro Wainzinger @xevix.bsky.social · 27/01/2025

Not only did it pick new encodings, but seems like it improved on general encoding strategy. Original was using straight up plain data in many cases it seems.

110 0

Alejandro Wainzinger @xevix.bsky.social · 01/01/2025

Excited to see materialized CTEs officially on the roadmap in Clickhouse. It's a difficult feature to implement but will be a gamechanger in performance of complex queries using temp tables as a workaround right now.

010 0

Alejandro Wainzinger @xevix.bsky.social · 16/12/2024

The most useful AI feature in iOS right now for me is Safari summaries. Much more likely to read a full post if the summary sounds interesting. This post echoed some of my own concerns with being a generalist benn.substack.com/p/is-this-a-...

000 0

Alejandro Wainzinger @xevix.bsky.social · 13/12/2024

Asking Qwen 2.5 coder (via Open WebUI/ollama) about PEG parsers, while having screen wide enough to see all columns of taxi dataset in DuckDB, on M4 Max MBP mirrored to Vision Pro feels like the future is almost here.

020 0

Alejandro Wainzinger @xevix.bsky.social · 12/12/2024

5k ultra wide virtual display in Vision Pro is certainly one way to look at code. Wow.

020 0

Alejandro Wainzinger @xevix.bsky.social · 10/12/2024

Wow, the Netherlands post international shipping is fast! Some nice DuckDB swag just in time for the holidays. 🎉

010 0