About this project

Emburk is a Rust-native bulk data loader that aims to preserve the pluggable and parallel design of Embulk while removing the dependency on the JVM. It is currently in the bootstrap stage, focusing on establishing a stable execution core and compatibility contracts. Key characteristics include: - Rust-native core: Designed to run without a JVM, with plugins progressively reimplemented in Rust. - Pluggable architecture: Supports statically linked built-ins and external plugins via a versioned, isolated protocol. - Migration path: Provides optional JVM and JRuby hosts to bridge compatibility with existing Embulk plugins. - Focus on correctness: Prioritizes transactions, cleanup, cancellation, and resume behavior over raw speed. - Evidence-based compatibility: Claims support only for pinned versions that pass differential tests against the original Embulk. Near-term goals include delivering a native File-to-File vertical slice (supporting CSV, JSON, and compression) and implementing a Java compatibility host for Maven-style plugins.