Drop #423 (2024-02-19): If It Walks Like A…

SQL Workbench; Tabulator; DuckDB Remote Parquet Two Ways

You'll have to head over to the main Drop site for the first two sections, as we're just covering the third one here to take advantage of better code formatting, and for a Nue teaser.

DuckDB Remote Parquet Two Ways

In the CLI version of DuckDB we can connect to remote data sources and even perform SQL ops across all of them. When DuckDB was ported to WASM, it retained those capabilities, though you won't be saving out database files from your browser just yet.

This means we can treat a bunch of remote parquet files as individual tables in one browser-homed database and perform operations on them.

We're going to do with some smaller, real-world datasets of mine to show how this works. And, we'll do it with the raw DuckDB WASM library, and Observable's super spiffy version of it, just to show the (slight) differences.

In each, we'll display the results in a Tabulator table, and we'll walk through the entire process, breaking down the steps in Glow-formatted code chunks.

Most of the expository is in the vanilla example, with Observable-specific commentary in the Observable example.

The examples are broken up this way so you can just view the source of each, independently, and not get confused about which imports (etc.) refer to which example.