Slide 6
Slide 6 text
7/21/23, 8:04 AM libSQL: Taking SQLite to the Moon
https://compileralchemy.substack.com/p/f5ee57d9-3127-45d5-88c8-4d55034d3c1e 2/41
A fork of SQLite, if it can dare be forked, is pure madness if they intend to be serious about it.
They would need to provide the same guarantees as SQLite. Stable, reliable, and dependable
software means the introduction of careful, slow-moving changes. Innovation might sometimes
be a radical change of components to unleash more powers, which in turn spells out instability, a
huge put-off. The team behind the fork must be competent. The SQLite team with decades of
experience, codebase intimacy and field specialization already pushes improvements to the limit.
To be able to propose alternatives, you must have a good notion of databases to craft an even
better product. And wisdom. If a fork of a car is a plane, there are two drawbacks: car users will
most probably abandon it as it’s no longer a car. And, since it’s a plane, it will be compared to
planes, it must be up to the level and satisfy plane users. The leadership team must be able to
find a purpose to the project, fuel motivation for a long time to come and make sure great people
remain around even if nobody adopts the project, a hard challenge to answer.
And indeed, many projects do not fork SQLite. They add components over SQLite and package it
as an improved project. Forks of SQLite exist. They come and crumble. But, sometimes ago an
unusual fork appeared: libSQL. This post discusses libSQL and it’s jaw-dropping team, the
technical challenges they overcame, the adoption strategy they put in place, the production
validation scheme they concocted, the ongoing development, the coma of future roadmap
checkpoints, how features resonate more than fully with a cloud-dominated industry and why it’s
the grandest, noblest and most soundly engineered forks of sqlite of all time.