Skip to content

v1.2.0

Choose a tag to compare

@v1lleneuve v1lleneuve released this 11 Jun 02:03
· 1 commit to main since this release

The Performance & Precision Release

The v1.2.0 release is a major architectural upgrade focusing on high-performance streaming and aggressive hardware-level cache management.

Important

Industrial-Grade Performance
This update introduces "True Streaming" logic. sys-shred now uses par_bridge to parallelize discovery and destruction simultaneously, allowing it to handle massive directory trees (millions of files) with a fixed, ultra-low memory footprint.

Key Enhancements

  • True Streaming Architecture: Removed the legacy "collect-then-process" model. Files are now shredded as they are found in the directory walk, dramatically reducing latency and RAM usage.
  • OS-Specific Cache Bypassing:
    • macOS: Now uses F_NOCACHE to explicitly instruct the kernel to bypass the page cache, ensuring data is sent directly to the hardware.
    • Linux: Implemented posix_fadvise (DONTNEED) to proactively flush overwritten data from the system cache, preventing "lazy writes" from lingering in volatile memory.
  • High-Precision Progress: A new dual-pass discovery engine ensures that even with the new streaming model, the progress bar remains 100% accurate without sacrificing performance.