Engineering blog

Notes from the compression engine

Byte-exact lossless compression, the best-of-N racing dispatcher, and the +1 byte no-expansion floor. The Orchestrator races xz, zstd, brotli and your data's SRD math and keeps the smallest verified output - never larger than the best standard codec (worst case +1 byte). Reproducible numbers, no overclaiming.

2026 - 6 min read

Structure-aware, byte-exact, never-worse: structured data, byte-exact, and where it falls back

The difference between entropy-coding compressors (gzip, brotli, xz) and structure-aware compression. On data with discoverable structure the win can be large; on incompressible data Pedulli falls back to identity with a +1 byte floor. Every restore is byte-exact and reproducible on request.

Read the post ->

2026 - 8 min read

How Pedulli's best-of-N racing dispatcher actually works

Inside the router: how a file is routed by magic-bytes + extension + an entropy probe, how parallel codecs race and pick the smallest output, and the 1-byte container format that guarantees no expansion.

Read the post ->

2026 - 5 min read

The +1 byte no-expansion floor - and why every codec should have one

Measured on 1 MiB of random bytes: gzip adds ~341 B of framing, brotli ~5 B, xz ~424 B, 7-Zip ~658 B. Pedulli adds 1 byte. A guarantee about Pedulli's own output - never-worse than input + 1 byte - and the design choice behind it.

Read the post ->

See the real number on your own data

Drop a file in the browser trial, or check pricing for the API.