How we made Haskell search strings as fast as Rust

Re:How we made Haskell search strings as fast as Rust by Tarean on Wednesday March 13, 2019 @07:29PM (#19385049)

I wondered about this and poked at the internals of ByteString (suitable for ffi) and Text. Turns out Text could allow ffi by changing a single function (newByteArray to newAlignedPinnedByteArray).

Presumably creating lots of small strings is a typical use case for Text so losing compacting gc hurts. ByteString on the other hand is more used with binary data and is designed for ffi use.

[ Parent | on HN ]

Re:How we made Haskell search strings as fast as Rust by whateveracct on Thursday March 14, 2019 @03:07AM (#19387066)

Hmm ByteString has the opposite issue and solved it with ShortByteString. Seems like PinnedText could be the analogue.

[ Parent | on HN ]