the 90x figure is on Go source for apples to apples against CGO bound tree-sitter.
replyyour use case is not one i designed for although yeah maybe the readme has some sections too close. the only external scanner missing atm is norg. now that i know your use case i can probably think of a way to close it
So your benchmarks are primarily just “how fast is go’s c interop” rather than any algorithmic improvement on tree-sitter?
replyEdit: yep, you are just calling a c function in a loop. So your no-op benchmark is just the time it takes for cgo to function. I would not be able to get any perf benefits from e.g. rust