News for hackers. Stuff that matters.
Re:Limbo: A complete rewrite of SQLite in Rust by encoderer on Tuesday December 10, 2024 @02:10PM (#42380238)
Even without that, it’s helpful. It means there is less (no?) undefined behavior that you will need to emulate to maintain compatibility. You can just follow the spec.
Re:Limbo: A complete rewrite of SQLite in Rust by dotancohen on Tuesday December 10, 2024 @02:29PM (#42380477)
If you can not run the test suite, then how do you know that you properly followed the spec? And did so securely? And in a performant manner? Even for edge cases? On obscure hardware, filesystems, and OSes? Even if the power cuts out? Or the cable to the hard drive (transactions)? Even if a stray cosmic ray flips a bit?

By the way, SQLite itself does not meet one of these criteria. Know which one? ))

Re:Limbo: A complete rewrite of SQLite in Rust by encoderer on Tuesday December 10, 2024 @02:33PM (#42380512)
I’m not sure what your point is? Yes, it would be better if they would run their tests against your fork. But they won’t. Still, it’s better for the fork writer that they exist.