Considering the great lengths the SQLite source goes through to validate correctness (code that rides in critical airplane systems gets no free lunch), I'd be immediately concerned a "SQLite in ______" lacks remotely the same degree of vigor in ensure implementation correctness. The overhead in using cgo feels like a small price to not worry "SQLite in Go" will remain maintained to the same degree the base SQLite project is.
They have `0 errors out of 928271 tests`. The code is mostly auto-generated if I recall, so it's essentially just using translated C from the original SQLite project.
