yaegi VS starlark-go

Yes. There are long standing feature requests for (e.g.) the reflect package that simply don't get done because they'd break this assumption and/or force further indirection in hot paths to support "no code generation at runtime, ever".

Packages like Yaegi (that offers an interpreted Go REPL) have "know limitations, won't be addressed" also because of these assumptions.

https://github.com/golang/go/issues/4146

https://github.com/golang/go/issues/16522

https://github.com/traefik/yaegi?tab=readme-ov-file#limitati...

There is always https://github.com/traefik/yaegi - a Go interpreter written to make it easy to write plugins.

I think https://github.com/traefik/yaegi comes as close as using the python interpreter in you CLI, but for Go

There's a go interpreter: https://github.com/traefik/yaegi It could run programs without compiling them, but there're some limitations.

I'd recommend looking into a different approach for plugins such as hashicorp/go-plugin (which uses multiple process PIDs and RPC communication between them) or traefik/yaegi (which implements a Go-compatible scripting language that can be interpreted at runtime and which still supports most Go modules).

Why do we need a language like Rust when we have Rust. Why not just create a Rust interpreter. (There's such an interpreter for Go, BTW, https://github.com/traefik/yaegi )

A cool solution I saw was Traefik's yaegi module. They basically created an interpreted scripting language with Go compatible syntax (turning Go into an interpreted, not compiled, language). I haven't tried this but it sounds like it brings the better parts of dynamic languages like Python's plugin support to Go - plugin writers can still write "Go" code, which can load dynamically.

Starlark in go https://github.com/google/starlark-go is a great way to combine the best of both, the ease of use of Python and the simplicity of go.

I have been building a platform for deploying internal web applications using this approach https://github.com/claceio/clace. Use Starlark to configure the application, the platform itself is built in go.

The big unknown is your task definition: what does user-defined logic look like? If you're expecting go code, that's gonna need some cleverness because of the compiled nature of it. There's a node runtime implemented in go if you want to provide sandboxed javascript (check the source of k6.io, it's the main one I know that uses it). If you want to provide building blocks and let them compose them, starlark might be a good choice.

Another one of those (with broader language support) is the Starlark language, which has a Go implementation: https://github.com/google/starlark-go

Every config format gets as complex to be touring complete in the end. We had similar problems and eventually got rid of that complexity and switched to starlark (the bazel config language), was a huge benefit for the tools. https://github.com/google/starlark-go "Starlark is a dialect of Python intended for use as a configuration language. Like Python, it is an untyped dynamic language with high-level data types, first-class functions with lexical scope, and garbage collection."

Direct link to the Go implementation of Starlark: https://github.com/google/starlark-go

For a Python-like syntax, https://github.com/google/starlark-go is the language used in Babel. It's very mature, but since it is used in a massive mature project with a specific purpose, it doesn't move fast or drift from the spec of its Java-based sibling. It doesn't have exception try except blocks or some other features you might expect, but for short extension logic, it might be exactly what you want with the stability you can depend upon.

Context: https://github.com/google/starlark-go/blob/cfacd890221418a2dc2c736f7b5e3476c38709b1/starlark/int_posix64.go

Thus envd is more like Dockerfile, while it uses a simplified python dialect starlark https://github.com/google/starlark-go as the build language.

Use Go (mostly for starlark-go)