How GZIP compression works
summary
Explains that GZIP uses DEFLATE = LZ77 + Huffman, compresses data in blocks with three modes (no compression, fixed Huffman, dynamic Huffman), and uses a 32 KB sliding window for LZ77 matches. Compares implementations: GNU gzip vs 7‑Zip vs zopfli — zopfli spends ~100× more time to gain ~5%, so use offline precompression for deploys. Shows preprocessing wins: transposing JSON and ordering attributes can cut sizes (one experiment reports ~17%).
AI-written summary · about
chapters (6)
- 0:00Start
- 12:15HUFFMAN CODING
- 12:50BLOCKS
- 14:36IMPLEMENTATIONS
- 19:53TRANSPOSING JSON
- 21:24XML/HTML ATTRIBUTES ORDER
related talks
43mzip Code: Unpacking Data Compression
46mNot Only Gzip
42mHoney, I Shrunk the Bits. Or how modern HTTP compression works.
32mBeyond Zstandard: The Future of Data Compression
47mWeb Compression: How to (Re)Take Control?
24mknock-knock-who-there-file-compression-TALK-FINAL-4.tar.trz.bz2.gz
18mWhy Lossless Data Compression Is Important and Fun
09mCompressing the Library of Babel!
