soa-derive VS styx

For those interested, Odin and Jai (I think) implement this at the language level. D can do this with its metaprogramming. Rust can also do it, but with procedural macros (i.e. the token stream is parsed and transformed prior to semantic analysis).

I'm looking for a way to derive an Arrow RecordBatch from an array of structs. So, it should act (almost) just like soa-derive. Has anybody done this?

> I'd love to see the codegen differences between the applications

There are two applications, dynamic calls and dynamic casts.

Dynamic casts to final classes dont require to check the whole inheritance chain. Recently done this in styx [0]. The gain may appear marginal, e.g 3 or 4 dereferences saved but in programs based on OOP you can easily have *Billions* of dynamic casts saved.

[0]: https://gitlab.com/styx-lang/styx/-/commit/62c48e004d5485d4f....

The @SOA attribute.

I've experienced that today, in my compiler. The language has tuples but also an internal comma expression (i.e not part of the grammar, just of the compiler API). As expected it was 100% possible to remove it and refact using tuples.

The problem described in this article is solved in Styx. During a pre-semantic pass the this parameter is added to member functions that are not static hence member functions are type- correct. Then, from a binary point of view, this is passed in a register.

Once upon a time I've profiled my software. It's a compiler for a hobby language called Styx.

Is there a reason why your table define the index at the tail of the dedicated row ? (my generator)

Since several years I work on a hobby programming language called Styx. The compiler was initially written in D. D sources were translated at the end of 2021 and finally Styx sources replaced the D ones on the 24 of January 2022, when Styx became "self-hosted".