The latest release of duckdb/duckdb is v2.0.0. DuckDB is an analytical SQL database that runs inside the application process. It can store and query data without a separate database server, and it also analyzes CSV and Parquet files directly. However, the provided FACTS contain no release notes or change items for this version, so what actually changed in v2.0.0 is not stated.
The core of this project is that it is a database embedded into applications and used for analytical work. According to the README, it supports not only basic SQL but also nested correlated subqueries, window functions, collations, and complex types such as arrays, structs, and maps. There is a standalone CLI plus Python, R, Java, and Wasm clients, and it integrates with packages such as pandas and dplyr. The implementation language is C++ and the license is MIT. In other words, it offers a configuration that fits cases where you want to use SQL analytics across various environments without operating a separate server.
The specific improvements in this release and any performance comparison data cannot be confirmed from the provided FACTS. The repository's features and supported environments are described, but there is nothing that ties them to the changes or performance improvements in v2.0.0. To judge what this version update means, the release notes need to be checked.