CREATE TEMP TABLE and getvariable() are my favorite DuckDB features. Thanks to this DuckDB is a natural fit for interactive data dashboards. Temporary tables allow you to define datasets once and reuse them across a dashboard. Variables make defining dynamic data filters intuitive.

I would probably never have built a data visualisation tool if I knew before how much math it takes. Somehow I imagined charting libraries handle it all. But here I am coming up with algorithms and magic numbers to properly format chart legends and labels no matter what data you throw at Shaper.

Why did I decide to build an embedded analytics product? I was setting up a data platform for a client. The scale and complexity of their data was trivial. But they needed to be able to embed data dashboards into their application. And the solution had to run on premise. 1/2

I am all-in on @duckdb.org. With Shaper you can build data dashboards all in SQL. Powered by DuckDB. Why DuckDB? Because of how much flexibility its simple design enables. My customers use DuckDB extensions to query Postgres, Google Sheets and even trigger webhooks. And we are just getting started.

To scale a business you have to eventually work with other people. Yes, AI can help each person to be more efficient. Someone still needs to take responsiblity for the results. That's what you pay people for. Also means functional org structures reflect who is responsible.

My problem with React is not JSX. I like it. But the way data updates is so complex to reason about and it often feels super inefficient too. Doesn't seem like the solution we should settle with to build all our UIs.

Why is rendering PDFs still such a mess? Browser all print differently. Even two headless chromes don't give the same result. Anyone has a solution they are happy with?

Didn't expect @taleshape.bsky.social to be mentioned in the NATS newsletter! But yes, Shaper makes heavy use of NATS to replicate @duckdb.org across multiple nodes, schedule tasks, buffer data ingestion and more.

This feedback is the reason I am building @taleshape.bsky.social: I want to create dashboards by simply writing SQL. And I want a tool that feels great to self-host. #duckdb #dataengineering #datanalytics #sql

"Garbage in, garbage out" is not the actual problem companies are having with AI. We are mostly stuck at "nothing in, nothing out".

Are you using Bluesky and want to stay on top of what’s happening? Are you curious how you can use Shaper to pull data from APIs and build interactive dashboards, all in a single tool and with just SQL? I created a little demo project: taleshape.com/blog/build-y... #databs #duckdb

Thanks a lot @nanovms.bsky.social for featuring us! That's a great walk through video!

We have a first Youtube walk through to get started with Shaper! And it's even me that recorded it. Thanks a lot @nanovms.bsky.social! Really appreciate the shout out 🙏 Great to see you can also run Shaper as unikernel.

DuckDB is awesome and it’s a great tool to explore and transform data. But DuckDB doesn’t help you visualize and share data with others. That's why I built Shaper: taleshape.com/blog/turn-yo... #duckdb #databs #dataengineering

I have $10 Anthropic credits that are expiring this week. Now I need to find some work Claude is good at. Looks like it's time to add new features instead of working on my actual priorities.

Working on a fun demo project 👀 #databs #duckdb #dataengineering

Can you tell that these 2 SQL queries do the same thing? DuckDB comes with lots of conveniences for writing SQL. I don't want to go back to thinking about trailing commas ever again,

I just shipped the biggest update since I started building Shaper. This is why I am excited about Shaper's new Tasks feature: taleshape.com/blog/why-i-a... #buildinpublic #dataengineering #databs #duckdb

It's easy to feel like you need to know more to do your best. But somehow successful people are also super knowledgable. They are just more selective with their input. Reading a few good books is more valuable than keeping up with the latest trends.

How do you balance between input and output? How do you decide that you learned enough and it's time to process all the knowledge and turn it into action? I often notice that the most successful people lean much more towards output than most of us.

anyone uses vanilla-extract for css in js? i am not unhappy with tailwind, but I like the idea of keeping bundle size small. i also feel like i forgot how to even write vanilla css at this point.

Anyone else finds AI much more helpful for frontend than backend work? Or is it because of how much more popular React is over Golang?

The product I am building is only possible because of tech we didn’t have 10 years ago. Not even talking about AI. And 10 years from now all of this will be completely outdated. It’s all about the value we can create now, and how well we adapt to changes.

Everyone told me how hard it is to make money with open source software. That's why I had to give it a try myself. Starting today, Shaper is not only free to use, but completely open source: taleshape.com/blog/shaper-... #buildinpublic #databs #dataengineering

Shaper now supports a cool DuckDB feature: Temporary Tables. Temporary tables allow you to pre-compute data to reuse within a single dashboard. Seeing how nicely temporary tables integrate into Shaper makes me feel great about the bet to make Shaper dashboards essentially just SQL scripts.

Shaper now supports public dashboard sharing and I added a public demo dashboard to the landing page. I was looking forward to this feature to easily share examples and use Shaper to play around with public datasets. taleshape.com #dataviz #buildinpublic

When you spend so much time on data analytics, you use SQL to calculate your visa due date

Taleshape now has a blog. And I just published the first post! Check it out if you like to know more about what Embedded Analytics is and how Shaper can help you build analytics functionality into your product. The post also comes with a small live demo 👀 taleshape.com/blog/what-is...

when chatgpt does a better job pitching @taleshape.bsky.social than me

Eight months ago I set out on my journey as solo developer and started building Shaper because I was frustrated with how complex it is to create data dashboards that you can embed into your application:

How to build a business with open source? -Enterprise Support, Managed Services, Consulting, Monitoring -Proprietery Extensions for Auth, Clustering, Analytics, Monitoring and Integrations - Cloud Services - Updates, Monitoring, Analytics, APIs, Security, Backups

My fav therapist just dropped some great content specifically for us software devs 🙏 www.youtube.com/watch?v=XW-0...

I was thinking that frontend has come a long way building my React app over the last months. I was excited about all the available components so I don't have to build everything from scratch. Now I have to update dependencies. And everything breaks. Without clear migration path. What do I do now?

Is it just me or does Claude still very little about DuckDB? I feel like it has very little idea about the available functions and definitely not about DuckDB specific Syntax. Anyone has a solution for this?

There is only one reason I am thinking about getting a new laptop with more than 16GB of RAM: TypeScript LSP Server.

Why do I feel like MacOS and Windows are both becoming more buggy over the years, and Linux on the desktop is becoming more and more stable?

Do I still need wifi if my phone gives me this?

i honestly just want to use react router for my next project because of these retro graphics on their landing page

I just learned about Malloy. Such an intriguing alternative to SQL for not only querying, but also modeling data. And they have a concept to tag values which can be used to render data. That's even more elegant to what I try to do in SQL using types. www.malloydata.dev

Is there a reason why we separate our data warehouse from our application database, or is it only a performance optimisation because no one has built a data store that unifies transactional and analytical workloads well enough?

Is there a reason why we separate our data warehouse from our application database, or is it only a performance optimisation because no one has built a data store that unifies transactional and analytical workloads well enough?

Can't deny how smart Cursor is, but I miss the speed and simplicity of nvim. Cursor makes me want to buy a big screen, a mouse and a powerful computer. Maybe that's been their business model all along 🫠

Regular reminder that if you register with this feed and add the emoji 📊 to your posts, they will appear in the feed, reaching a wider dataviz audience and making Bluesky better for the dataviz community.

Just starting to realize how many events monitoring systems produce: Pulling stats every minute doesn't sound like a wild throughput: 60*24=1440 events per day. But you are not pulling one metric. If you have 100 timeseries/min that's 144k events/day. 52 million/year. And that's just one server.

this update to my website was overdue

Teaser of what it looks like to use the data analytics tool I am releasing soon to build interactive dashboards in plain SQL. Custom SQL types define how the result of each query is rendered. It's all just SQL. And LLMs are pretty good at SQL.

AI sucks at generating example data. But it's great coder. Worked the first try: "I have a table in Postgres with the following columns: ... Write a script that generates data for a hundred rows and gives me an insert sql statement I can run."

Side effects in React are a pain. I am loading data in useEffect and it's loading multiple times. I wrapped all functions in useCallback and I know that in dev mode React calls useEffect twice just to annoy you. And still I don't get it right.

TIL about SQL UNNEST( ) Generate multiple rows out of one without UNION or cross joins.

Note to self: Stop using "FROM scratch" Docker images. The simplicity is always tempting, but it sucks when TLS root certs are missing, you don't have a /tmp directory and you cannot sh into the box to debug it.