Show HN: Query local or remote CSV, Parquet or Arrow data in the browser

sql.quacking.cloud

2 points·tobilg··3 comments

Absolutely cool but the dominant data format on the web today is JSON and you'd do even better with an object-relational query language (like N1QL)

Thanks for the feedback! DuckDB can handle JSON data via the JSON extension, but this hasn't landed yet in the DuckDB-WASM package itself which is the basis of the adhoc querying features.

The intended audience of this little tool is more in the direction of data engineers or data analysts, which search for a quick and easy way to analyze datasets up to a few million records...

I think a lot of those people would benefit from the object-relational (somewhat realized in PostgreSQL) model.

My take is that a lot of those people struggle with joins and in the pure relational model you end up having to create complicated queries with joins that could be expressed more naturally with objects with nested structures.

I had the hardest time trying to sell that idea to some private equity backed investors who were buying up "boxes-and-lines" data analysis tools and relational DB engines because they were 110% committed to columnar query execution engines though.

That said, I did sweep through the DuckDB docs and found a lot of great stuff in there, it really is rich in features for data analysis.

Reply on news.ycombinator.com