Hi everyone,

My name is Ivan and I’d like to introduce Arx, an experimental programming
language being developed with support from Open Science Labs, and some
feedback from the Arrow community would be very appreciated.

A core idea behind Arx is to use Apache Arrow as the native foundation for
its data-oriented runtime and data types. The current runtime already uses
Arrow C++ for arrays, tensors, DataFrames/Series, and RecordBatches, while
the language-level support is still being expanded.

At a high level, Arx uses ASTx as its AST representation, while IRx handles
semantic analysis and lowers supported ASTx nodes to LLVM IR using
llvmlite. On the native side, we have a C++ runtime built on Arrow C++,
exposed through a C ABI that the generated LLVM code calls into.

We already have the foundations in place for Arrow-backed arrays, tensors,
DataFrames/Series, and RecordBatches, including Arrow C Data
interoperability for arrays and IPC/PyArrow interoperability for
RecordBatches. There is still quite a lot to build and improve.

With support from Open Science Labs, we also had two Google Summer of Code
contributors working on the project, which helped us make significant
progress on the Arrow integration.

At this stage, I’d really value feedback from the Arrow community,
especially around the architecture, use cases worth exploring, missing
pieces we should prioritize, and ways Arx could fit into or complement the
broader Arrow ecosystem.

We’re also looking to grow the community around the project, so if anyone
is interested in contributing, collaborating, or knows people or
organizations that might be interested in helping us move it forward, I’d
be very happy to connect.

Project: https://github.com/arxlang/arx

Thanks!

Ivan Ogasawara

https://ivanogasawara.com

[email protected]


Schedule a Calendly meeting with me here:

https://calendly.com/ivan-ogasawara


Open Science Labs

Founder & Executive Director

https://opensciencelabs.org

[email protected]

[email protected]


[image: github] <https://github.com/xmnlab>[image: linkedin]
<https://www.linkedin.com/in/ivan-ogasawara/>[image:
https://iogasawara.substack.com/] <https://iogasawara.substack.com/>

Reply via email to