Port to Go Lan?

[k105la]

https://gocv.io/writing-code/hello-video/ https://godoc.org/gonum.org/v1/plot https://www.gonum.org/post/introtogonum/

[ngiangre]

@tesla809 you said you were learning GO?

[tesla809]

Great questions!

td;lr: I'm down for it!

Pros: -If we go sever model, it would be faster than Python and can allow for scalable operations -If we go edge computing with WebAssembly (operating on phone or browser), we can use TinyGo implementation and compile to WebAssembly.

• Easier than doing it in Rust or any other language. GoLang is essentially a replacement for Python for systems programming.
• Of lesser note: I get to apply the GoLang I am learning to a project.

Cons:

• Extra time to write it a new language.
• Fewer libraries for machine learning applications (seems less efficient) in GoLang.
• Less high-level ML libraries in Go than Python.

Overall I am for doing it. Sounds like fun and it seems like the pros outweigh the cons. We have better advantages is the distribution model (servers & clients). Related to libraries, if we go with Microservices route, this might not be an issue. We can implement one thing in one language, another in another.

Thoughts?

See: https://www.reddit.com/r/golang/comments/anbjin/is_there_a_future_in_machine_learning_in_go/efscc0p/

[tesla809]

@ngiangre Yup. Going strong. Should be decent at it by the end of the week. Then ... (cue ominous music) Rust.

[ngiangre]

Ok let's do it so we can have everything client side! Let me know if you'd like to team up somehow

[k105la]

Yes it would take longer to write but I was wondering if the long term benefits out weights that. Also go-ipfs would be so useful!

[ngiangre]

I think the long term benefits are greater. If we can do everything client side that goes with our philosophy including decentralization and no data on external servers (user privacy but also low data/server maintenance for us). Just means we gotta figure out a way to get that from a user's gaia