Fastalp: lossless floating-point compression in pure Rust

Hey everyone, Over the past few weeks I have been working on fastalp, an adaptive lossless floating-point compression library implemented in pure Rust. If you deal with time series, telemetry, or columnar datasets, you have probably noticed that raw IEEE 754 floats do not compress very well with general-purpose byte compressors (zstd, lz4) or integer bitpackers due to the exponent/mantissa bit layout. The original ALP paper from ACM SIGMOD 2024 (by Azim Afroozeh et al., which DuckDB and FastL...