ADASS posters are displayed all week
When
Theme: Other creative topics in astronomical software
The Astronomical Table Serialisation System (ATSS) comprises a data modelthat represents tabular data along with the abstractions needed for serialization and deserialization across diverse data formats such as CSV, VOTable or FITS.
Drawing inspiration from the Serde.rs framework, ATSS shares conceptual similarities with the "format-neutral" core of STIL within TOPCAT. One of ATSS's distinctive features is its capability to discern three distinct data representations: a conventional in-memory representation; an also conventional ASCII representation; a potential representation at the storage byte level. The byte-level storage representation offers compression opportunities while preserving the FITS BINTABLE property of a consistent byte count per row.
The QATSS tool has been developed to perform queries on files in any format supported by, and implemened for, ATSS. Its capabilities encompass row filtering, column selection, position-based queries (including but not limited to cone, multi-cone, and MOCs queries), index-based queries, and multi-threading, among others. Notably, QATSS is currently in active production within VizieR to query large catalogs stored internally in a specialized CDS format.
This presentation will delve into the motivations driving the development of ATSS and its distinctive attributes. It will also showcase the QATSS tool along with comparative benchmarks. The broader ecosystem will be presented, including Rust implementations of essential components like an expression evaluator, a file-stored binary-search tree code, and diverse file format implementations including VOTable. The presentation will conclude by exploring the potential implementation of the parquet file format within ATSS, potentially simulating a 'hipscat' feature.