I've been using nushell as my daily driver for about six months now, and wanted to show a few simple examples of why I'm enjoying it so much. I think it's a breath of fresh air compared to most shells.

Why a new Shell?

In case you've never heard of it before, nushell is a, well, new shell ;) bash has been the dominant shell for as long as I can remember, though zsh have their fair share of devotees. fish is the only recent example I can think of as a "challenger" shell. fish gained enough traction that it's supported by tooling such as Python virtualenv (which only has integrations out of the box for a handful of shells). I think fish is popular because it had some slightly saner defaults out of the box, was easier to "customize" with flashy prompts (which can make your shell SUPER slow to init), and had a saner scripting language than bash. But it still retained a lot of the historical baggage from POSIX shells.

Nushell challenges two common assumptions about shells and asks "what if things were different?"

1. POSIX compliance is a non-goal.
2. Many standard tools from GNU coreutils/base system (e.g. ls and du) are replaced by builtins.
3. All nushell "native" utilities produce and consume structured data rather than text by default.

By dropping the goal of POSIX compliance, nushell frees itself from decades of baggage. This means you get a scripting language that feels a lot more like Rust. You'll actually get errors by default when you try to do something stupid, unlike most shells which will happily proceed, usually doing something even more stupid. Maybe treating undefined variables as empty string make sense in the 1970s, but that's almost never helpful.

nushell also takes a relatively unique approach to utilities. When you type something like ls or ps in nushell, this is handled by a shell builtin! It's just Rust code baked into the shell rather than calling out to GNU coreutils or whatever your base system includes. This means that whether you type ps on FreeBSD, Debian, or macOS, you'll get the same behavior!

I can already hear some readers thinking "doesn't this just massively bloat the shell?" No, not really. The code for these is for less than that of the typical GNU utility, because nushell actually (IMO) embraces UNIX philosophy even better than the original utilities. They are all extremely minimal and work with other builtins. For example, there are no sorting flags for ls, and no format/unit flags for du.

The reason that nushell can take this approach is because they challenge the notion that "text is the universal API." You can't meaningfully manipulate text without lots of lossy heuristics. But you can do this for structured data! I admit I'm a bit of a chaos monkey, so I love to see a project taking a rare new approach in space where nothing has fundamentally changed since the 1970s.

Okay, enough about philosophy... here are a few examples of some shell pipelines I found delightful.

Elastic snapshot status

First up: I do a lot of work with Elasticsearch during $DAYJOB. One workflow I have to do fairly often is spin up a cluster and restore from a snapshot. The Elasticsearch API is great... for programs. But I have a hard time grokking hundreds of lines of JSON. Here's an example of a pipeline I built which culls the JSON response down to just the section I care about.

http get "http://localhost:9200/myindex/_recovery"
  | get restored-indexname
  | get shards
  | get index
  | get size

In case it's not obvious, the http command makes HTTP requests. This is another nushell builtin that is an excellent alternative to curl. It's not as feature-rich (curl has a few decades of head start), but it brings something new to the table: it understands from the response that the content is JSON, and converts it into structured data! Everything in this pipeline is nushell builtins. And it'll work on any OS that nushell supports. Even Windows! That's wild!

Pro tip: you can press option+enter to add a new line when typing in the shell.

Disk usage in bytes

Here's an example I hinted at earlier. If you type help du (to get built-in docs), you won't find any flags for changing the units. But you can do it using formatters like so:

du path/to/bigfile.bin | format filesize B apparent

The du command always shows human-readable units by default. Which I very much appreciate! And did you notice apparent at the end there? Well, the version of du you'd find with a typical Linux distro doesn't exactly lie to you, but it withholds some very important information. The physical size occupied on disk is not necessarily the same as how large the file (in an abstract platonic sense) actually is.

There are a bunch of reasons for this, but the most impactful one is compressed filesystems. If I ask Linux du how large a file is in an OpenZFS dataset, it will report the physical size by default, which may be a few hundred megabytes when the file is really multiple gigabytes. Not necessarily helpful.

Anyways, the nushell builtin always gives you columns for both physical and apparent. So you can't ignore the fact that these sizes are often different. I like that!

Some other helpful bits for switching

If you want to give nushell a try, they have some great documentation. Read the basics, but also check out their specific pages on, e.g. coming from bash.

Finally, here are two more that tripped me up at first.

• If you want to get the PID of your process, use $nu.pid instead of $$.
• To access environment variables, you need to be explicit and go through $env. On the plus side, you can now explicitly differentiate from environment variables.