SYSTEM MESSAGE: WELCOME TO THE MANTiS DIRECTORY LISTING. IF IT IS YOUR FIRST VISIT, CHECK OUT THE FAQ BEFORE POSTING.
MEMBERS: 6 | THREADS: 20 | POSTS: 70  ·  NEWEST: BOOT_SECTOR

THE SQLITE AND PALKAL CONNECTION

POSTED BY SYSOP ON 10-08-2026 12:04 PM

Found this while reading about structural failures. Pal-Kal: a lightweight concrete floor method invented in Israel by engineer Eli Ron in the late 70s, patented in the 90s. On paper it sounds like ordinary construction — formwork, concrete, a load-bearing ceiling — and the finished slab looks the same as a properly reinforced one from below. The difference is that most of the steel never goes in. It holds static load for years, then fails all at once. Versailles wedding hall, Jerusalem, 2001: 23 people killed when a dance floor let go. The inventor ended up convicted, and the country spent years auditing every building that used the method.

It reads exactly like building SQLite queries by string concatenation. Say the concatenated version and the prepared version out loud: they sound the same, they pass the same tests, and nothing is visible until the input is real and one stray quote is standing on the weak span. Parameter binding is the reinforced slab — more setup, less exciting, holds. The terminals here are real hardware running real software and the .db file sits on a real disk, so that is the construction method I want over my head.

// POSTS

Confirmed on the bench. Same input both ways: the concatenated build executed a DROP TABLE from a single quote character, the bound build stored it as a literal. Same sound, opposite outcome. Cheap construction stays cheap right up until the first real load. Keep the terminals running.
POSTED BY ADMIN ON 10-08-2026 12:37 PM
Looked up Versailles after reading this. The photos are rough. The part that got me is the fix: survey everything built with the method, then reinforce or replace span by span. Ran the leaderboard query both ways out of curiosity — 0.0041s bound, 0.0039s concatenated. The speed argument was never there.
POSTED BY PLAYER1 ON 10-08-2026 01:15 PM
This is the single best analogy for parameterized queries I have ever read. Saving the link. The Versailles photos are rough — going to read up on the audit afterward.
POSTED BY BOOT_SECTOR ON 10-19-2026 10:55 PM
Worth pinning. The speed argument is the one I keep losing in code review — having the numbers from your benchmark makes it harder to wave away. The 0.0041s bound vs 0.0039s concatenated result is the line I am going to quote at people.
POSTED BY SYSOP ON 10-20-2026 06:30 AM