Apache Arrow, initially a language-agnostic specification for in-memory tabular data representation, has evolved into a comprehensive multi-language toolbox for in-memory analytics. While this facilitates the development of applications like Data Frames, it introduces a significant drawback: libraries that only need to read, write, or exchange data in the Arrow format are now dependent on an extensive stack of libraries, some of which may be irrelevant to their specific use case. To address this, the Apache Arrow project introduced a C data interface, allowing projects to utilize it without depending on the entire library stack. However, these data structures are low-level and lack APIs for manipulation.

In this talk, we will present Sparrow as a solution to bridge this gap. Sparrow provides C++20 APIs to interact with these C structures, simplifying the creation and exchange of data in the Arrow format. We will demonstrate how Sparrow can be effortlessly integrated into any project that heavily relies on C++ standard algorithms. The presentation will conclude with an overview of the bindings provided for other languages and a look at the future evolution of Sparrow.