Final Thesis: Optimizing Internal Data Representation in Jayvee

Abstract: Jayvee is a simple language for describing data pipelines. The execution of these pipelines necessitates the handling of tabular data. Previously, Jayvee utilized TypeScript’s data structures to represent such data. This thesis develops a new table implementation for the Jayvee interpreter. We present the architectural design and implementation of such a prototype. It uses the Polars library to adhere to the Apache Arrow specification. Additionally, the library sqlite-loader-lib, written in Rust, is integrated into this architecture, to accelerate the export of tables. The evaluation demonstrates, that the new implementation has the potential to increase the Jayvee interpreter’s maximum input size from 475 Megabyte above 2:5 Gigabyte and its processing speed by a factor of between 3:60 and 18:22.

Keywords: JValue, Engineering Thesis

PDF: Bachelor Thesis

Reference: Jonas Zeltner. Optimizing internal data representation in Jayvee. Bachelor Thesis. Friedrich-Alexander-Universität Erlangen-Nürnberg: 2024.

Video: https://widgets.wp.com/likes/master.html?ver=20250109#ver=20250109&n=1