It said I clicked on water a few times when I had not... Maybe it was a canal? Very cool.

This happened to me too (on mobile), every tap said it was in the water

Oof I think that's a bug. Let me investigate!

I've moved from python/polars/pandas to DuckDB and have not looked back

Same. Almost every time I would use its streaming interfaces in Python, it would STILL materialize everything into memory. That was like 6 months ago. Maybe streaming interfaces actually work, but I found them to be leaky abstractions that required a ton of hand holding to make sure they didn't build a bunch of memory pressure, if you're lucky enough to even have a way to do it.

For example, last time I used it, you couldn't do NDJSON streaming scans from S3 (looks like fixed with PR #26563).

SQL is a terrible api for data transformation.

https://duckdb.org/docs/lts/sql/statements/pivot#limitations

vs

https://docs.pola.rs/api/python/stable/reference/dataframe/a...

even for just in-memory quick data analysis?

spinning up a venv with duckdb: 5 minutes

using sql everywhere; never having to remember dataframe syntax: timeless

what is the visualisation strategy here? I find post-hoc grouping and plotting in pandas pretty satisfying.

Check out malloyyo on github

Yep, and chdb as well. Said this before, chdb's DatStore is a pretty neat pandas replacement too.

The article mentions a dataset of names given to people in the USA. This is a fun name explorer WASM based on that data. 811 people in the USA have named their baby 'Diesel': https://docs.malloydata.dev/malloyyo-babynames/name_explorer...

you should checkout https://github.com/malloydata/malloyyo - it's great at what you trying to do.

Thanks for sharing, and thanks for making the video you shared. A couple thoughts. Is PostGIS king in this area? I've been liking duckDB as it does not require a server. Is there a good alternative to PostGIS in the duckDB world? Maybe just plain GeoParquet files read in through duckDB?

In the example in your video are any special GEOMETRY functions being used in the underlying SQL? Or, could your data just have been in plain postgres?

A similar product GeoSQL is Malloy which puts a semantic layer on top of your data for better LLM understanding. Malloyyo gives you an MCP server for precise and auditable interaction with your data.

Thanks for sharing all this. Where is the submission page on claude / openAI to submit MCP servers to their directories?

on claude you'll need an enterprise and team account here is their post about it https://claude.com/blog/observability-for-developers-buildin... on chat you can find it in the openai platform https://platform.openai.com/apps-manage - need a paid account too

Question: I have an MCP server that is working well with Auth (Google Only), running on vercel. What benefit would I get from running it on manufact?

On monetizing my MCP... How do the different MCP "Stores" handle this? Do some take a cut? Or are they agnostic? Does manufact help with monetization?

Thanks for the comment! Vercel is a generic cloud provider, so you won’t get any of the MCP specific features (listed in the post) and development experience - monetization is still not regulated by the protocol, so people would pay for your product before using your MCP, then you can charge based on subscription or usage as usual!

Do you use MCP-handler for your MCP ?

This is awesome. Is your code open source? It would be cool to make a textbook for SQL in this format.

Yes, code here: https://github.com/nlothian/llm-sql-benchmark

There is a V2 branch with some minor fixes etc I'm working on.

Can it be run as a static site(github pages) ?

It is completely static - everything is done on the client side.

It's WASM plus client-side React components in a static site.

I'm hosting of Cloudflare Pages (without their serverside function hosting), so yes GH pages will work fine.

If you are a data scientist or do anything with data... duckdb is like a swiss army knife. So many great ways it can help your workflow. The original video from CMU in 2020 [1] is a classic. Minutes 3-8 present a good argument for adding duckdb to your data cleaning/processing workflow.

And if you want to add a semantic layer on top of data, Malloy [2] is my favorite so far (it has duckdb built in):

[1]: https://www.youtube.com/watch?v=PFUZlNQIndo [2]: https://docs.malloydata.dev/documentation/

Thank you for the recommendation on that video! I've already adopted to using DuckDB for my ad-hoc analytics work but I didn't know the background

Nice work. Have you blogged about how you built it?