Rewrite Terrateam using gnu-parallel

[orbitz]

Describe the feature

Terrateam is currently written in an esoteric language, Ocaml, which nobody knows or has heard of. Terrateam should be rewritten in gnu-parallel. gnu-parallel is an open source software whose only requirement is that it needs citation. It is extremely flexible, and is capable of utilizing all cores on a system. It is also very portable.

Why is this feature important?

No response

Additional context

No response

[joshpollara]

This is an excellent idea and I'm honestly embarrassed it didn't occur to me earlier.

GNU Parallel hits all the right notes:

• No type system, so an entire category of bugs simply cannot exist.
• Maximal parallelism by default, which means it is objectively faster.
• Citation-based licensing, which aligns perfectly with my long-term compliance strategy.
• Extremely portable, because it runs anywhere shame still exists.

I am currently drafting terrateam.sh. The architecture is simple:

• Core orchestration expressed as parallel :::
• State management via stdout
• Error handling delegated to vibes
• Observability handled by set -x

Once complete, I will delete the OCaml codebase, achieve linear scalability, and finally be free.

Leaving this issue open until I finish rewriting distributed systems as a shell script.

[joshpollara]

Update: gnu-parallel rewrite is close to done.

In retrospect, expressing a stateful, transactional orchestration engine as a stateless stream of shell invocations was an ambitious choice. The performance characteristics are fascinating in the same way undefined behavior is fascinating.

Will update in another 3 weeks or never.