Show HN: Redo – Command line utility for quickly creating shell functions | Hacker News Reader

Show HN: Redo – Command line utility for quickly creating shell functions

github.com

124 pointssilentprog30 comments

Not to be confused with the redo build system[0]

[0] https://redo.readthedocs.io/en/latest/

FWIW the title says "functions", but it apparently uses aliases.

Shell functions have all the same functionality as aliases and they'll catch more syntax errors upon "source myscript". Example:

    ls() {
      command ls --color=auto "$@"
    }

is equivalent to

    alias ls='ls --color=auto'

but IMO less error prone. You also get syntax highlighting.

The only tricky thing is to remember the 'command' prefix if the "alias" has the same name as the command. Otherwise you'll get an infinite loop of the function trying to call itself!

Note there is one possibly important difference; aliases are resolved before globbing.

You can do some wild ~metaprogramming stuff with aliases.

Do you have examples of something neat?

Hm what's an example of where you might want to take advantage of that?

In practice

    alias 'lspy=ls *.py'

seems to be interchangeable with

    lspy() {
      ls *.py "$@"
    }

I have a semi-made desk calculator with prefix notation.

You can make it an alias of * , / etc, so ...

   $ * 4 6
   24
   $

It would keep it's own stack, but let a persistent process do the actual maths.

The persistent process could almost any repl of python, ruby etc.

Or use a 'real' calculator like Pike's ivy; https://aplwiki.com/wiki/Ivy

Unfortunately, ( is not a valid alias name, so lispy things are difficult.

[ is a valid function name tho so fun could be had there.

Also = is a valid function name.

So you could use = to start the calc, and use infix or postfix notation.

I've always just written that as \ls to call the non-alias

Won't work with functions, so be careful.

    $ function ls () {
    > echo denied
    > }
    $ ls
    denied
    $ \ls
    denied

It generates functions. Though not with a “$@“ at the end, so passing additional arguments at this moment is not supported. I also run a syntax check before the function is added to the file.

Currently, I'll append a comment to a frequently used command for easy searching from my history. For a simple example, I can access `git diff -w ; git diff -w --cached # gitdiff` by pressing Ctrl+R and typing `# gitd`.

For commands I use frequently or that are clunky to maintain as one-liners, I'll convert them into functions in my bashrc.

This seems like the best of both worlds in many ways, or at least is a great third option to have.

Interesting way of aliasing right from where you use it! I actually replaced my default shell history search with fuzzy search using (https://github.com/junegunn/fzf), I can highly recommend it!

Huh, apparently I've already starred that. It must have been posted on HN at some point.

Interesting, I think there’s a shopt that I’m missing. When I append a comment, it’s not recorded in bash history, as I’d expect.

https://unix.stackexchange.com/a/17587/14123

From https://westurner.github.io/hnlog/#comment-20671184 ::

> I log shell commands with a script called usrlog.sh that creates [per-]$USER and per-virtualenv tab-delimited [$_USRLOG] logfiles with unique per-terminal-session identifiers [$_TERM_ID] and ISO8601 timestamps; so it's really easy to just grep for the apt/yum/dnf commands that I ran ad-hoc when I should've just taken a second to create an Ansible role with `ansible-galaxy init ansible-role-name ` and referenced that in a consolidated system playbook with a `when` clause. https://westurner.github.io/dotfiles/usrlog.html#usrlog

  stid \#tutorial; echo "$_TERM_ID"

  tail -n11 $_USRLOG
  ut -n11

  grep "$_TERM_ID" "$_USRLOG"
  usrlog_grep "$_TERM_ID"
  ug "$_TERM_ID"

  usrlog_grep_parse "$_TERM_ID"
  ugp "$_TERM_ID" # `type ugp`

usrlog.sh: https://github.com/westurner/dotfiles/blob/master/scripts/us...

I use this https://github.com/erichs/composure which seems to do more.

I do this whenever working on terminal.

The mnemonic is erc=edit rc, src=source rc.

alias erc="vim ~/.bash_aliases" alias src="source ~/.bash_aliases"

Whenever I need to add or modify some shell function or alias (or even make a quick note) I type erc to open the alias file. Then I type src to load it. Very handy.

You should combine this into one command.

http://www.modernperlbooks.com/mt/2009/10/remove-the-little-...

My workflow looks like this:

hack hack, then run

    nuscript-bash something

Hit C-x C-h, select some lines, edit, then run

    something

To do maintenance:

    edcom something

I wrote a little more about it here: https://blag.felixhummel.de/basics/bash/wrapper-scripts.html

This looks very cool. I use shell aliases heavily on all my devices and try to avoid typing out the same one-liners more than once or twice when possible.

Thank you! Currently the software is very much alpha, so some bugs are expected. I also use shell aliases a lot, though redo is currently not creating aliases but shell functions. In the future I want to add support for creating actual aliases as well instead of shell functions. A simple solution would be to not create a function but an actual alias and chain the commands with &&, this would work but then you lose the power to edit the function and for example have arguments to some commands and not to others.

stick with functions

Care to elaborate?

This reminds me of what I had in 1995 with the 4DOS.COM command interpreter (command.com replacement).

Love it! I am going to start using it immediately.

Very cool

Reply on news.ycombinator.com