@[email protected]

TIL you can turn any website into an editable canvas with just a simple command in the console: document.designMode = 'on'

The Geo for Good Summit session recordings are available now. Check out the 5-hour workshop on the Earth Engine Python API and geemap 👇 YouTube playlist: youtube.com/playlist?list=… #geospatial #EarthEngine #geemap #GeoforGood23

@boristane JSON-RPC servers admitting they aren't RESTful: htmx.org/essays/how-did…

wait, am I reading this right? and the comments agree? I'm truly confused so to delete an item you make a POST to /resource/id/delete ? or do you make a POST to /resource/id with body { action: 'delete' } what is this?

I produced my first book trailer 🙃 find it here: youtu.be/xubQ0lsiV44 And if you watch until the end, there may be a surprise gift 🤔

Worried that rain 🌧 will ruin your trip? @sensibleweather gives travelers the confidence to book that next trip by offering weather-guarantee protection. It also provides weather insights using climate analytics. bit.ly/3FhBg0a #NSFSBIR ☔ 📸: Sensible Weather

Great updates and additional content in 2nd edition. Project based learning on a variety of fun topics (🐧!). Deploy your apps and adapt them to your own interests! Stoked to be a part of it, great work @tylerjrichards !

Today i'm releasing 🥟 st-paywall, a @streamlit component that lets you build subscription Streamlit apps. Just run pip install st-paywall, add your Google+Stripe keys, add add_auth() to your app, and you've got a subscription Streamlit! github.com/tylerjrichards…

@minimaxir I like the simpleaichat cli so far. Very little friction having a random terminal open with a running chat

The Problem With LangChain

Larger than RAM group bys have landed in @DuckDB's master branch! This means that now every operator in DuckDB can handle larger than memory intermediates. So the max data a single node can handle is the size of your SSD, not RAM! #issuecomment-1620424908" target="_blank" rel="nofollow noopener">github.com/duckdb/duckdb/…

Just standup things

New post: I read a really great book recently, "Naming Things", that explains why picking the right abstractions in development is so hard. vickiboykis.com/2023/06/29/nam…

💧☀️ Awesome summary!

@FranckPachot @duckdb @PostgreSQL Great examples of other ways to write this query! I’m getting worse performance with the window function on Postgres even with indices on the 100k example. (DuckDB scan stayed constant) Fair nit-pick on the “CTE” definition

@GarsBar35Plus @duckdb @PostgreSQL -- Real life tables have primary keys. Try: alter table students add primary key (id); alter table exams add primary key (sid,id); vacuum analyze; -- I see no CTE but correlated vs non-correlated subquery -- Analytics in SQL should use window functions 👉dbfiddle.uk/wLZ4H496

CTE vs Subquery: Who Gives a 🦆uck?! New blog post comparing @duckdb and @PostgreSQL performance + their combined power. Heavily inspired by de-correlated queries duckdb blog and referenced paper home.gerardbentley.com/blog/cte-subqu…

@duckdb @PostgreSQL related duckdb blog: duckdb.org/2023/05/26/cor… related paper: cs.emis.de/LNI/Proceeding… github repo for demo code: github.com/gerardrbentley…

@ghalimi Are you looking for historical climatology at each location / date or historical observations? Looks like a fun exploration!

After a bit of work, we managed to get time series for weekly average temperatures in 2012, but only for 1,286 stores out of 5,182 stores. Bottomline: we need a better dataset of historical temperatures.

We are now ready to match stores using temperatures. For that, we need a dataset of daily temperatures, and the Global Historical Climatology will be perfect for that. ncei.noaa.gov/products/land-…

@__AlexMonahan__ @TheRockstarDBA This is spicy! I’m trying out the first query from the paper with 100k students. Preliminary results: duck db scanning pg: 1.4 s duck db cached: 0.26 s Pg: still running

@TheRockstarDBA Yes! Everything except the reading of the original data is processed in DuckDB, so this algorithm would be used!

With DuckDB, even correlated subqueries are lightning fast! One benchmark runs in 0.06 seconds with DuckDB vs. over 48 *hours* in Postgres! Think of correlated subqueries as parameterized subqueries! Use aggregates for filtering, ex: show me all orders from my top 10 customers.

@andfanilo @streamlit @langchain It’s always the edge cases 😉

Ok, I admit, building a @streamlit app to query and summarize multiple 100 pages (public) PDF documents with @langchain in less than 5mn is a great feeling🤩 Now into the devilish details, like what if those documents are in multiple languages, tables in PDF, images in pages

@zahnster we've got backends for frontends and now frontends for frontends! what a time to be alive!

This article is ridiculously good. bradfrost.com/blog/post/lets…

New blog post by @Maxxen_: PostGEESE? Introducing The DuckDB [Geo]Spatial Extension - duckdb.org/2023/04/28/spa… #GIS