Hi HN — I’ve been building Otters, a DataFrame library for Go designed to bring a Pandas-like workflow to native Go without requiring Python, JVM, or external runtimes.
GitHub: https://github.com/datumbrain/otters
I started this because most Go DataFrame libraries I tried were either abandoned, hard to use, or didn’t feel idiomatic to Go. I wanted something that:
• works fully in-process (no cluster / Spark dependency)
• uses native Go types instead of generic interface blobs
• supports expressive chained operations
• stays readable and explicit like Go code should
• can scale to large datasets in memory pipelines
Design goals were:
• simplicity over magic
• type safety over dynamic guessing
• composability for real data pipelines
It’s still early but functional, and I’d genuinely appreciate feedback from people who’ve built data tooling or worked on columnar engines / query planners. Critiques welcome — API design, performance, architecture, anything.
GitHub: https://github.com/datumbrain/otters
I started this because most Go DataFrame libraries I tried were either abandoned, hard to use, or didn’t feel idiomatic to Go. I wanted something that:
• works fully in-process (no cluster / Spark dependency)
• uses native Go types instead of generic interface blobs
• supports expressive chained operations
• stays readable and explicit like Go code should
• can scale to large datasets in memory pipelines
Design goals were:
• simplicity over magic
• type safety over dynamic guessing
• composability for real data pipelines
It’s still early but functional, and I’d genuinely appreciate feedback from people who’ve built data tooling or worked on columnar engines / query planners. Critiques welcome — API design, performance, architecture, anything.