sixten VS FlatBuffers

Immutable data structures don't necessarily require more memory: they can avoid deep copies. They are also automatically thread safe without expensive (slow) locking mechanisms. They also don't necessarily reduce cache locality. The reduced cache locality in the case of Haskell (I think) mainly comes from the representation of objects in its implementation (improved STG) which uses extensive boxing and jumps that hinder both spatial and temporal locality (require review/comment from GHC/Computer Architecture experts, take it with a grain of salt). Objects can be much more efficiently represented if not for the need to implement lazy (call-by-need) semantics. See sixten and futhark for examples.

[Sixten](https://github.com/ollef/sixten): functional programming with unboxed data by default.

I can't answer this well and don't know of any resources, but I have seen it before in the parser for sixten:

    https://github.com/ollef/sixten/blob/60d46eee20abd62599badea85774a9365c81af45/src/Frontend/Parse.hs#L458

Sixten is a language that allows precise control over memory layout of algebraic data types.

For something more peripherally related, check out Sixten. Its focus is on using unboxed value representations, which is in spirit close to what you are proposing, and some of its ideas might be good inspiration.

I don't work at Cloudflare but follow their work and occasionally work on performance sensitive projects.

If I had to guess, they looked at the landscape a bit like I do and regarded Cap'n Proto, flatbuffers, SBE, etc. as being in one category apart from other data formats like Avro, protobuf, and the like.

So once you're committed to record'ish shaped (rather than columnar like Parquet) data that has an upfront parse time of zero (nominally, there could be marshalling if you transmogrify the field values on read), the list gets pretty short.

https://capnproto.org/news/2014-06-17-capnproto-flatbuffers-... goes into some of the trade-offs here.

Cap'n Proto was originally made for https://sandstorm.io/. That work (which Kenton has presumably done at Cloudflare since he's been employed there) eventually turned into Cloudflare workers.

Another consideration: https://github.com/google/flatbuffers/issues/2#issuecomment-...

Much more in the discussion of this particular issue onGitHub: flatbuffers:Reverse engineering #4258

Related but not Rust-specific: FlatBuffers, Cap'n Proto.

If json files are a concern because of space, you can always look into something like protobuffers or flatbuffers. But whatever you use, you should try to find a solution where you don't have to think about the actual serialization/deserialization of your objects, and can just concentrate on the data.

Or, as an alternative to protobuffers, there's also flatbuffers, which is lighter weight and needs less memory: https://google.github.io/flatbuffers/

As the title suggests, I am considering using Flatbuffers as a way to parse sensor data that has been stored in local datafiles. The project language is python.