April 8, 2026
He SQL’d the penguin, chaos ensued
Pgit: I Imported the Linux Kernel into PostgreSQL
He crammed 20 years of Linux into a database — and the comments exploded
TLDR: A developer loaded 20 years of Linux code history into a PostgreSQL database with pgit in about two hours, then surfaced quirky stats. Comments split between title nitpicks, AI-ghostwriting jabs, performance jokes, and real interest—impressive tech, but the drama stole the show.
A bold dev just stuffed two decades of Linux’s change history into a PostgreSQL database using a tool called pgit — about 1.4 million changes in roughly two hours — then ran SQL queries to find curiosities like only seven swear-filled messages (from two people), one mega-bug that 665 fixes point to, and a file system that took 13 years to land. Translation for non-nerds: they moved the Linux code’s entire diary into a spreadsheet-on-steroids and started asking juicy questions.
But the real fireworks? The comments. One camp unleashed the title police, insisting it should say “Git history,” not “the kernel,” complete with nerdy regex flair. Another burst out laughing after misreading the headline, joking it sounded like a wild fix for Linux performance troubles. And then came the drama bomb: a reader called the post “GPT-ish,” accusing it of AI ghostwriting and begging the author to dial it back. Cue a split between “this is very cool” and “I can’t read this tone,” while a curious crowd chimed in with “Fossil uses SQLite” nostalgia and “I might try pgit now.”
The vibe: impressed by the stunt, divided by the words. The tech worked; the discourse worked harder. If you enjoy a little data wizardry with your semantic warfare, this post delivered.
Key Points
- •The full Linux kernel Git history (1,428,882 commits) was imported into pgit, a PostgreSQL-backed Git-like system using pg-xpatch for delta compression.
- •The import completed in 2 hours 0 minutes 48 seconds on a Hetzner dedicated server with AMD EPYC 7401P, 512 GB RAM, and RAID 0 SSDs.
- •Repository metrics included 24,384,844 file versions, 3,089,589 unique blobs, and 171,525 unique paths, with total data size of 2.7 GB (vs. 1.95 GB with git gc --aggressive).
- •Environment setup used Ubuntu 24.04 LTS, Docker for pg-xpatch, and system tuning (CPU governor performance, mitigations off, THP disabled, noatime).
- •Post-import queries over SQL-exposed history surfaced facts like 7 expletive instances in messages, 665 bug fixes referencing one commit, and a filesystem merge spanning 13 years.