Add Zip library

[ernestognw]

🧐 Motivation

For L2s that publish data to the L1, it comes handy to have a library to compress/decompress data, reducing L1 publication costs

📝 Details

[Amxx]

While we agree on implementing some lossless compression logic, there are many questions to discuss:

• I'm pretty sure we want to implement decompression, for usecases where the (call)data is compressed offchain, passed compressed, and decompressed onchain. But do we also what to implement compression? Are there any usecases for doing compression onchain?

• Solady implements 2 algorithm:

  • LZ77 based on FastLZ
    • the ariya/FastLZ refers to an npm package that hasn't been updated in 9 years, and that doesn't install. I tried to find other LZ77 implementation, but I couldn't not find anything compatible. Most are low quality
  • A calldata specific algorithm that doesn't appear to be standard.

I'm wondering if we want to implement either of these, or if we should favor something more standard.

IMO we should go for an algorithm:

• that has is standard
• that is optimised for fast/cheap decompression
• for which there are packages/libraries in various languge (among which javascript/typescript)

[Amxx]

According to ChatGPT, we should probably implement LZ4, LZO or Snappy

Algorithm	Decompression Speed	Memory Use	Compression Ratio	Simplicity & Ecosystem
LZ4	Extremely fast (GB/s)	Minimal	Moderate	Simple, widely supported
LZO	Very fast	Minimal	Moderate	Very lightweight, older but robust
Snappy	Very fast (~500 MB/s+)	Minimal	Lower	Easy to implement; Google-supported
LZAV	Fast (~2–4 GB/s)	Minimal, safe	Slightly better than LZ4	Newer, portable C library
Iguana	Extremely fast (6–13 GB/s)	Requires SIMD	Comparable-ish	Experimental, hardware‑specific

Algorithm	Ease of Decompression Implementation	Standardization and Ecosystem Support
LZO	✅ Easiest – minimal logic, ANSI C, zero extra memory	Good legacy support but fewer modern ports
LZ4	🟡 Simple but requires framing/token logic for full compliance	✅ Highly Standard – kernel, filesystems, multi-language libraries
LZAV	❌ Most complex – limited implementations, no framing standard	❌ Very limited adoption & bindings
Snappy	🟢 Simple, robust C/C++ API with error checking; moderate framing complexity	✅ Broad Ecosystem – widely used across major big-data and storage systems