v0.3.0
Pre-release
Pre-release
High-Performance Parallelism & Forensic Standards
The v0.3.0 release is a major technical milestone, introducing multi-threaded execution and industry-standard erasure algorithms. This version transforms sys-shred into a high-performance suite capable of handling massive data volumes with forensic-grade precision.
Tip
Performance & Multi-threading
This release introduces rayon for parallel file processing. On multi-core systems, recursive shredding operations are now up to 10x faster by distributing the cryptographic workload across all available CPU threads.
Key Enhancements
- Parallel Destruction Engine: Fully integrated
rayonto enable concurrent shredding of multiple files. This significantly reduces execution time for large directory trees. - Standard Erasure Algorithms: Added support for international data destruction standards including US DoD 5220.22-M (3-pass) and the Gutmann (35-pass) method via the
--methodflag. - Hardware Verification: Introduced the
--verifyflag, which performs a post-pass read-back of data blocks to ensure physical persistence and bypass hardware-level remapping failures. - Safe Execution (Dry-Run): Added a non-destructive
--dry-runmode to preview operations, critical for validating recursive targets and exclusion patterns before commit. - Pattern-Based Exclusions: Implemented
--excludesupport usingglobset, allowing users to skip specific files or directories (e.g.,.git/,*.log) during recursive operations.
Technical Implementation
- Thread-Safe Core: Refactored the
ShredderandOverwritercomponents to be fully thread-safe and compatible with parallel iterators. - Unified Progress UI: Re-engineered the terminal interface to provide a coherent, aggregate progress view for multi-threaded operations.
- Hardened Directory Cleanup: Optimized the bottom-up cleanup logic to gracefully handle partially excluded directory structures.
- Documentation Overhaul: Completely updated
README.mdwith detailed technical specifications, safety warnings, and advanced usage examples.