Interesting since in some ways, as he points out, it's in direct competition with DataFrames for use cases, but he gives it a very positive treatment and shows how they can work together using advantages of standard SQL along with processing power of DataFrames.
When someone gives fair opinions on something that directly competes were their own work, you should take their opinion very seriously. It’s an excellent quality in a person, and shows they’re more focused on the problem than their ego.
Quite a while ago, when duckdb was just a duckling, I wrote an R package that supported direct manipulation of R dataframes using SQL.[1] duckdb was the engine for this.
The approach was never as fast as data.table but did approach the speed of dplyr for more complex queries.
Life had other things in store for me and I haven’t touched this library for a while now.
At the time there was no Julia connector for duckdb, but now that there is, I’d like to try this approach in that language.
Interesting since in some ways, as he points out, it's in direct competition with DataFrames for use cases, but he gives it a very positive treatment and shows how they can work together using advantages of standard SQL along with processing power of DataFrames.