Frequently used Terminal commands

Created 4/23/2008, updated 8/26/20236 minute read

Apparently this "meme" has been spreading rampantly for a while now, but I only just found out about it here, here and here.

The idea is to see what commands appear most frequently in your Bash history using this command:

history | awk '{a[$2]++}END{for(i in a){print a[i] " " i}}' | sort -rn | head

My results:

4598 git
641 cd
585 ls
438 rake
284 less
232 script/console
188 cap
148 ll
140 perl
118 irb

I was pretty surprised that Capistrano (cap) and perl rank so highly. But this is because every time I do a Rails deployment I end up doing a series of about 10 cap invocations (staging deploy:check, staging deploy:update, staging deploy:migrate_all staging spec, staging deploy:restart, and then the same all over again for the production environment). And the abundance of perl usage is due to the fact that I used it pretty heavily to do a find-and-replace over thousands-of-pages dump of the old wiki.

Update: January 2011

Just for laughs, here are the results from running the command again, almost two years later:

5341 git
1005 cd
692 ls
369 less
220 open
163 bin/rspec
161 man
158 m
124 bundle
102 mv

Top 3 unchanged, rake completely gone (note that newcomer bin/rspec is probably responsible for that), open is surprisingly highly-ranked, cap gone (thank goodness), m (MacVim shortcut) and bundle are notable indices of how my software ecosystem has changed.

Update: July 2011

Fairly big changes over the last months:

5635 git
837 ack
626 less
487 cd
396 ls
212 ssh
121 m
121 clear
115 put
85 vim

Less cd as I am tending to stick in one project directory and working on it for longer. ssh is way up as I am working on other hosts a lot more.

Here’s the output on the host that I spend most time on:

1711 git
868 bundle
837 cd
419 less
367 ls
301 script/console
270 ruby
237 clear
214 ack
161 fg

All those bundle calls are from running things like bundle exec with things like spec, rake and mongrel_rails.

Update: February 2013

Another couple of years pass. Now that I’m on Zsh where history outputs only a handful of recent items by default, I’ve added a numerical param to it:

$ history -100000 | awk '{a[$2]++}END{for(i in a){print a[i] " " i}}' | sort -rn | head

Work machine:

6134 git
3524 ack
1142 less
676 ssh
429 cd
361 specr # wrapper around rspec
323 be    # this is an alias for `bundle exec`
318 rm
297 vim
270 bundle

Laptop:

1007 git
398 cd
312 less
274 ssh
124 ls
123 rm
114 vim
111 scp
109 ack
105 cat

No wonder people feel emotionally attached to Git, if it’s our most frequently used tool by such an enormous margin.

Update: October 2015

Laptop

3586 git  # Undefeated world champion.
672 less  # 2nd, up from 3rd.
345 vim   # Highest ever showing (3rd, up from 7th).
339 cd    # Down to 4th, from 2nd.
284 ls    # Steady 5th, despite Zsh hook that auto-lists on cd.
259 rm    # Steady 6th; I like to delete.
232 npm   # Goodbye Rails, hello JS ecosystem.
230 ag    # ack, dethroned
157 man   # Apparently RTFM-ing more than ever.
146 find

Work sandbox

3033 hg        # So sad.
1801 tbgs      # (Work) search tool.
931 less
623 ag         # Big code base = more searching.
511 arc
452 git
394 tgrep      # Alias wrapper around ag.
363 find|grep  # More searching: where/what is that file?
335 ls         # Less poking around, more time in (long-running) editor.
302 tgrepd     # Another alias wrapper around ag.

Update: November 2018

Note the need to tweak the command for my current Zsh configuration, otherwise history doesn’t show the full command list; fc -l 1 would also work:

history 0 $HISTSIZE | awk '{a[$2]++}END{for(i in a){print a[i] " " i}}' | sort -rn | head

The results:

824 git
173 less
166 cd
153 yarn
101 vim
81 man
75 brew
72 ls
65 rm
56 npm

Numbers are lower than before despite my $HISTSIZE being 100,000. It is probably because of my use of histignorealldups, which I am now questioning the usefulness of. I think histfindnodups is probably much closer to what I want.

Update: February 2019

Having tweaked my shell settings I can now report a more illustrative example. This one from my personal laptop:

4890 git
2034 yarn # using yarn to run scripts in projects
856 vim # open up mostly long-lived vim sessions
726 cd
535 less
480 exit # always opening and closing tmux splits
358 ls
299 ll # alias for `ls -laF`
250 o # alias for `git log --oneline` with a few other options added
222 man

Update: March 2020

I have been running with NO_HIST_IGNORE_ALL_DUPS, NO_HIST_IGNORE_DUPS and HIST_FIND_NO_DUPS for a while so have captured some more "exhaustive" data (ie. includes duplicates, so really counts all usage):

# Personal machine
12919 git
 5951 yarn      # mostly running scripts in projects
 2583 vim
 1782 cd
 1650 exit      # `exit` now beats `less`
 1094 less      # (last time, `less` beat `exit`)
  602 ls
  587 ll        # alias for `ls -laF`
  585 o         # alias for `git log --oneline`
  548 t         # "tmux" wrapper, knocking "man" out of 10th place


# Work machine
43359 git       # new job, so no more `hg`
 8015 yarn      # mostly running scripts in projects
 6239 cd
 5655 portool   # tool that wraps up a bunch of common work operations
 4314 exit
 3657 vim       # Vim sessions tend to be long-lived
 2888 less
 1927 npm
 1721 o         # alias for `git log --oneline`
 1395 ls

Observations:

Clearly doing more on work machine than personal machine.
I’m going to get rid of NO_HIST_IGNORE_DUPS because the consecutive duplicates are driving me up the wall when I hit "up" in the terminal (HIST_FIND_NO_DUPS stops you from seeing duplicates when you search, but just tapping "up" is not considered a search), so that will affect the numbers somewhat next time around.

Update: July 2021

# Personal machine
20379 git
6582 yarn
4602 vim       # Alias -> `nvim`
2986 cd
2927 exit
1624 less
1237 ll        # Alias -> `ls -laF`
963 o          # Alias -> `git oneline` -> `git --paginate log --pretty=format:'%C(auto)%h%Creset %s%C(auto)%d%Creset %C(magenta bold)(%cr)%Creset %C(cyan)<%aN>%Creset'`
866 ls
839 t          # Wrapper for `tmux`.

And here are some interesting entries from outside of the top-10; these produced with history 0 $HISTSIZE | awk '{a[$2]++}END{for(i in a){print a[i] " " i}}' | sort -rn | cat -n — the number on the left is the position within the list:

 12  721 man     # 721 instances of bad memory.
   ...
 17  459 tree    # A beautiful tool!
   ...
 18  347 brew    # Surprised this one is so high as I'm scared of running it.
   ...
 26  262 ssh     # Relatively rare on this machine nowadays.
   ...
 31  192 \vim    # 192 times I've bypassed my vim -> nvim alias
   ...
 37  166 sudo    # Again, surprised how rarely I do this.
   ...
 39  159 rg      # Would be more frequent, but I mostly search from inside Vim.
   ...
 44  135 hledger # A new toy.
   ...
 79  54 curl     # Fetching things, peeking at headers vs...
 95  37 wget     # Actually downloading things.
   ...
100  35 watch    # 100th place on the list!
   ...
212  10 prettier # Rarely run outside of `yarn`.
   ...
317  4 whoami    # Rare existential crises.
   ...
394  2 xattr     # The least popular tool I've used multiple times.