This is a cool project, I see it come up now and then. I've played with it in the past and have been impressed and I'm glad to see it's still being actively developed. I'll have to try it out again soon
It's become my daily driver for the light amounts of shell use I have (just derping around a filesystem and some data manipulation, not a sysadmin).
My only issue is that I haven't done any custom function/aliases/whatever for my more frequent locations because the docs on it seem intimidating and I'm not sure i'm looking in the right spot.
Being able to just open a document, say a json, and have it throw it in a table is soooo nice. I can then tweak it if need be, and save it back, or convert it to csv or whatever other format I need.
My only issue is that I haven't done any custom function/aliases/whatever for my more frequent locations
Don't reinvent this but get autojump or one of its many clones (it'll learn what your frequent locations are). Eventually combine with fzf to fuzzy search those, though I think something like zoxide does it all in one (but even then, you'll want fzf, it is just too handy to not use). I'm on PS so use ZLocation piped/sorted into fzf, that's just a oneliner and it should be straightforward to get something similar on NuShell. Once you've used that any environment which doesn't let you go to your most used locations (or history, or commands, or ...) in a matter of seconds and a couple of keystrokes feels seriously crippled. Because it is.
I've been playing with Nushell in past weeks and it's truly excellent.
I find it occupies a weird niche where I can do a bunch of data wrangling with it, and some basic exploration, but hit a ceiling where what I've done is less persistent than an ETL and less exploratory than a notebook.
I think a simple personal ETL system based on nu would be awesome, and likewise I want to add cell magic to Jupyter for it so that you can quickly get data into Pandas or data viz of some kind.
As someone who programs for a few years, I really dig their scripting language. I am a huge fan of small scripts but always found bash scripting scary as it is quite unique. Googling a specific thing and seeing some awk solution is just scary for me (like regex was before I understood it). That's not bad but I prefer having easily maintainable code for a small script.
Nu's principles and influence from modern languages makes it easier for me. But be careful about it not having a 1.0 release. Often times things can break, but their docs always point out breaking changes.
Bash is not simple. It’s unforgiving in many unique ways. And it’s difficult to master, unless you code in it every day, which is not a norm for most of us. And if you didn’t master it it’s dangerous, because it’s unforgiving. So it goes.
I've been using Nu as my default shell in Windows Terminal on my personal PC for a good year or so now. Can't recommend it enough. It integrates pretty well with windows PowerShell/CMD in that you can easily reference binaries outside of Nu's shell environment. Like PowerShell, Nu also has the concept of structured data primitives that can be passed between programs.
Between Nu and the new-ish winget package manager it almost feels like a proper linux distribution these days.
Is it a new kind of shell or a clone of powershell? What does it do better than pwsh? I'm genuinely curious because the Readme isn't really convincing.
15 comments
[ 2.7 ms ] story [ 44.4 ms ] threadhttps://news.ycombinator.com/item?id=35124056 (203 comments)
https://news.ycombinator.com/item?id=33419944 (247 comments)
My only issue is that I haven't done any custom function/aliases/whatever for my more frequent locations because the docs on it seem intimidating and I'm not sure i'm looking in the right spot.
Being able to just open a document, say a json, and have it throw it in a table is soooo nice. I can then tweak it if need be, and save it back, or convert it to csv or whatever other format I need.
Don't reinvent this but get autojump or one of its many clones (it'll learn what your frequent locations are). Eventually combine with fzf to fuzzy search those, though I think something like zoxide does it all in one (but even then, you'll want fzf, it is just too handy to not use). I'm on PS so use ZLocation piped/sorted into fzf, that's just a oneliner and it should be straightforward to get something similar on NuShell. Once you've used that any environment which doesn't let you go to your most used locations (or history, or commands, or ...) in a matter of seconds and a couple of keystrokes feels seriously crippled. Because it is.
I find it occupies a weird niche where I can do a bunch of data wrangling with it, and some basic exploration, but hit a ceiling where what I've done is less persistent than an ETL and less exploratory than a notebook.
I think a simple personal ETL system based on nu would be awesome, and likewise I want to add cell magic to Jupyter for it so that you can quickly get data into Pandas or data viz of some kind.
If nothing else, never having to use `jq` again is a huge QoL improvement.
Nu's principles and influence from modern languages makes it easier for me. But be careful about it not having a 1.0 release. Often times things can break, but their docs always point out breaking changes.
Between Nu and the new-ish winget package manager it almost feels like a proper linux distribution these days.
While using it I feel like working with a db like model. Also using bash commands works great and you learn to improve your parsing skills for free!