Brian
9 posts
River is getting better, with CTE, Cross DB joins, creating tables from queries, etc
Example use case: query users and their orders from a Postgres instance, and for some reason, dump them to a MongoDB/SQL Server instance.
Check this out:
Brian@superdispose
Federated queries are having a moment, Databricks Unity Catalog, Trino. They all solve the same problem: your data is split across systems, and you're tired of stitching it together with Python scripts and CSV exports. But here's the thing: those tools are built for the warehouse. They need clusters, catalogs, metastores, and a team to keep them alive. What if you just want to join a Postgres table with a Mongo collection from your terminal?
English
River is open source under MIT. The repo has a one-line curl install, prebuilt binaries for macOS, Linux, and Windows, and a full language reference covering every clause and expression in RiverQL.
If you've ever had data in two (or more) places and wished you could query them like one, this is for you.
Built with @rustlang and @ratatui_rs
github.com/bryanbill/rive…
English
The cross-database join is the headline feature. Say you have users in Postgres and orders in MongoDB.
River extracts keys from the build side, batches them, fetches matching rows from the probe side, and hash-joins the results. You don't think about drivers. You think about your data.
English
That's River. A single binary. No servers. No JVM. No catalog. You point it at your databases and start querying across them with one language, RiverQL, that works the same way whether you're talking to Postgres, MySQL, SQLite, MongoDB, or SQL Server (more adapters otw).
The engine figures out join strategies, pushes filters to the source, and merges results inline.
github.com/bryanbill/river
English
Federated queries are having a moment, Databricks Unity Catalog, Trino. They all solve the same problem: your data is split across systems, and you're tired of stitching it together with Python scripts and CSV exports.
But here's the thing: those tools are built for the warehouse. They need clusters, catalogs, metastores, and a team to keep them alive. What if you just want to join a Postgres table with a Mongo collection from your terminal?
English