NOTICE: This project is on hiatus and may not update for a while.

This project (yet to be officially named) is a collaboration with my friend Ejektaflex from the 2021 Wheel Reinvention Jam here on HMN. It allows you to explore a relational database without writing a single line of SQL.

I've always been frustrated by the experience of writing SQL queries - partially because of the syntax, but mostly because you can't actually see the data when you're writing them. Databases are great at storing information and making that information accessible. The way we access that data shouldn't be so cumbersome.

The tool is designed to make the data visible at every step of the way. Every node is a valid query. You can click on any node to view the query at that point, and then use that information to build the next step of the query. Queries that would normally require subqueries or common table expressions become trivial in this system, freeing you to think about the logic of the query itself.

This tool is simultaneously easier to use and more powerful than SQL queries. Or at least it will be when it's done. It's not done.

The project is currently on hiatus while Ejektaflex and I work on other important things and figure out some fundamental tech. We're also both pondering better designs for the nodes in the system, since the current ones stick way too closely to the design of SQL.

Some relevant links:

Recent Activity

Done. Super happy with how this turned out! Working on the full submission now. &db

I added aggregates, and @Ejektaflex added SQL pretty printing &db

Bigger example of the workflow. At each stage we can build up to a bigger query by looking at the data available to us. At the end we can easily make multiple "views" of the same data by making two Order nodes to sort the same data in different ways - and both Order options remain close at hand while we work forward from there. &db

Incrementally building database queries. Start with a table, see the data, tweak the data, repeat.

Every node in the graph produces a valid query. You can even build up multiple queries in the same "workspace" and run them on the fly.


such beauty, truly the future of databases &db