nordisk/znippy-zoomies

lgz — crate card

Reverse-engineered (from source — verify, don't trust the name)

• One-line pitch: A pure-Rust parallel gzip decompressor that parallelizes across DEFLATE full-flush boundaries with a zero-copy ring-slot pipeline.
• Does: Pipeline is Reader → Main(carry+split) → N Workers → Collector → Writer with ring-slot recycle. Locates full-flush markers (00 00 FF FF) where the LZ77 window resets so each segment is independently decompressible; streams from pigz/bgzf/Z_FULL_FLUSH parallelize fully, plain gzip output falls back to single-threaded flate2. Public API: decompress_gz, decompress_gz_capped, decompress_gz_into, decompress_gz_stream, deflate_offset, GzError, plus size-guard output_too_large/OUTPUT_TOO_LARGE. Modules: deflate_scan, chunk, parallel, speculative.
• Key deps: intra-workspace linflate (the actual DEFLATE core), flate2 (fallback + header handling), tar. Dev: flate2, miniz_oxide.
• Shape: lib + bin lgz (705 LoC); ~2.1k LoC. Feature timing. PARALLEL_THRESHOLD = 4 MB below which it stays single-threaded.
• Observed role in the workspace: Gzip decode backend for osm-katana (gzip-compressed OSM XML). Architecture explicitly mirrors lbzip2 and ljar; shares the DEFLATE core with them via linflate.

Fill in (one line each — only the human knows these)

• Primary consumer(s):
• Why it exists (vs the obvious alternative):
• Hard constraint (airgap / no-C-deps / no-FFI / pure-Rust / etc.):
• Release cadence & independence (rides a parent? own repo someday?):
• Status (production / prototype / experimental / stubbed):
• Must-NOT-change invariant:
• Biggest known gap or TODO:
• Anything a newcomer would get wrong about it: