This takes some getting used to. Suppose you type this:

mkdir /home/user/some-very-long-name

press C-c RET, then what you have on the prompt is this (I put in a vertical bar to represent the point/cursor):

mkdir /home/user/some-very-long-name ⋮ mkdir /home/user/some-very-long-name

so you can delete the second mkdir and type && cd or something.

Once you're used to this sort of feature, you'll always be looking for equivalents in other shells or feel handicapped. You could write your own Rosetta stone: a cheatsheet for equivalents in fish, oh-my-zsh, oh-my-bash, or whatever other shell you switch to in the future. All skills you learn in eshell are transferable.

The important thing is that you explore the most powerful features of whichever shell you're in so you know about them and know they're possible. That's how you grow your skill, and become able to exploit the potential of exotic shells like a riced-out Xonsh or Ambrevar's SLY-on-steroids. It builds up.

• C-c M-o – eshell-mark-output
• C-c C-p – jump to previous prompt
• C-M-l – eshell-show-output, almost the same, but jump to top of output
• C-c RET – more on this later
• C-c C-u – like C-u on a terminal obeying Bash hotkeys
• C-c M-d – eshell-toggle-direct-send – lets you say y to y/n prompts without pressing enter, I think

Some of my really nice custom bindings:

• C-M-j – my-insert-other-buffer-file-name
• C-c C-l – my-counsel-eshell-history
• <escape> – crux-switch-to-previous-buffer

You can type elisp expressions without outer parens, like

λ kbd "C-x f"

or

λ define-key

Try it! it returns

usage: define-key: (KEYMAP KEY DEF)

It's a bit like an alien fish. Oddly similar because in both shells, parens delineate nested function calls. The equivalent in Bash would be using grave accents, e.g. echo `concalc 1+2`. Try

λ echo (+ 1 2)
λ define-key $global-map (kbd "C-x t") #'crux-transpose-windows
λ list-timers

So cool! You really should try. If you don't have an Emacs open, open it and type M-x eshell. Type these commands out by hand.

More things to try:

λ concat a b
λ setq a 2
λ + a 3

Hear this:

Unlike regular system shells, Eshell never invokes kernel functions directly, such as ‘exec(3)’. Instead, it uses the Lisp functions available in the Emacs Lisp library. It does this by transforming the input line into a callable Lisp form.

[2024-04-08 Mon] That it just does lisp becomes obvious if you type a command that doesn't exist. Compare it in eshell versus shell.

To see the Lisp form that will be invoked, type:

λ eshell-parse-command "echo hello"
λ eshell-parse-command "echo hello; echo multiple words; setq a 2 b 3"

Also

λ which ls

One use I've for Eshell that's not possible in any other shell: after editing ~/.profile or ~/.xsessionrc or similar files, I can make it apply to the running emacs environment with simply

λ . ~/.profile

This is not possible in any other shell because they're external processes, whereas Eshell's "process" is the Emacs process.

May not seem important to source ~/.profile, but I feel there's something about the fact you can that proves there's no other sane shell choice. Emacs is supposed to be your HCI (human-computer interface). Why create roadblocks?

By the way, so you know, . .profile is is not the same as source .profile (in plain Bash, it is the same).

If you're looking for how to do something, try the info manual at C-h i d m Eshell <RET>. They tell you how to do for loops and whatever else you need.

• 2022-04-26
• Stupid tricks to quickly learn Eshell and Dired

Techno-optimism is unfashionable at the moment, but I suspect we still haven’t come close to realizing the potential of even the internet technology of the 1990s. When thousands of people converge on a topic, the collective knowledge far exceeds any one person, but our current interaction models don’t do a great job of synthesizing it. It’s a difficult problem, but it’s hard to imagine that in a hundred years we won’t have more effective ways to interact.

–www.greaterwrong.com/posts/8mjoPYdeESB7ZcZvB/observations-about-writing-and-commenting-on-the-internet

See also The Garden and the Stream

• Portal: Slipbox
• Incremental writing

• #design

Like AIC, you can rank models by WAIC. But a more interpretable measure is an Akaike weight. The weight for a model i in a set of m models is given by

$$w_i = \frac{ \exp{\frac12 \textrm{dWAIC}_i} }{ \sum^m_{j=1} \exp{\frac12 \textrm{dWAIC}_j} }$$

where dWAIC is the difference between each WAIC and the lowest WAIC, i.e. dWAIC = WAIC_i - WAIC_min.

New kid on the block, around 2020 it was the best (for which situations?).