Block a user
75dc0a6320
Modify alternative builder for alpine
30f5210ef5
Document alternative builder for scratch image
4d586c809e
Under 2 minutes, could probably start optimizing as I did for rust, but it's good enough for now
5bb2363eee
mmapped file slightly faster
eb2ed15e33
First julia impl, very slow
dfcc8562e6
ran cargo clippy
212e595a7e
ran cargo fmt
53ea542f36
update dependencies, create benches
d246c54cd9
removed polars example because it wouldn't compile anymore, even when updating to latest version. It also massively reduced the number of downloaded crates
2a89d061a0
Use FxHashMap for multi_threaded_smol.rs
7add8793a5
managed to get library version on par with the reference implementation by using memmap2. I'm understanding scoped threads a little better now... I think
b53212103b
Tried looking at what the performance would be if I read the data from a file where every line would have the same length ie. not having to read until eol. But despite not having to search the \n byte (or ; because every station name and temperatures are padded with null bytes) and having a fixed size array instead of vec this is slower (the normal read_until version is actually still just as fast, while the new one is 10x slower)