I find the idea neat, but I still struggle to see the benefits in my day to day job. The example doesn’t illustrate clearly what I cannot already do easily with bash, curl and jq:
I cannot for the life of me remember the myriad of filtering micro-DSLs that CLI tools like find and jq have.
I’m also curious to try a shell that isn’t only based on text. I suspect the Unix orthodoxy of “text only” is really just “stuff should compose easily.”
I’ve read about two new shell projects here this week (oilshell, and now this), in some ways I suppose I am glad that folks are still finding ways to hack on something that has been so core to our lives for so long.
Instead of treating all data as raw text, Nushell takes some inspiration from PowerShell and treats data as structured tables or objects.
This is most interesting thing for me about this project. I also liked reading about some of the stated goals of the project.
I find the idea neat, but I still struggle to see the benefits in my day to day job. The example doesn’t illustrate clearly what I cannot already do easily with bash, curl and jq:
curl https://jsonplaceholder.typicode.com/todos/ | jq '[.[].title]' > titles.jsonI cannot for the life of me remember the myriad of filtering micro-DSLs that CLI tools like
findandjqhave.I’m also curious to try a shell that isn’t only based on text. I suspect the Unix orthodoxy of “text only” is really just “stuff should compose easily.”
I’ve read about two new shell projects here this week (oilshell, and now this), in some ways I suppose I am glad that folks are still finding ways to hack on something that has been so core to our lives for so long.
This is most interesting thing for me about this project. I also liked reading about some of the stated goals of the project.
This is a really excellent point, we all depend on shells and haven’t seen much movement since things like Fish.
See also this thread from earlier this year: https://lobste.rs/s/fy5ivi/introducing_nushell
Great link. Thanks!