"Beautiful as poetry; ludicrous as science."

Efficient Editing With Vim

Translations of this article are available in German (Effiziente Textbearbeitung mit Vim), French (L’édition efficace avec vim), Italian (Vim e Gvim: usarli in modo efficiente, and Chinese (高效使用vim). Content © J. McPherson

“To me, vi is Zen.
To use vi is to practice zen.
Every command is a koan.
Profound to the user,
unintelligible to the uninitiated.
You discover truth every time you use it.”

This tutorial assumes a basic knowledge of vim — insert mode, command mode, loading and saving files, etc. It is intended to help vi novices develop their skills so that they can use vi efficiently.

In this tutorial, <C-X> means Ctrl-X — that is, hold down the Ctrl key and press X. You can get help on most of the commands used here by typing :help command in vim, where command is what you need help on.

Moving efficiently

Stay out of insert mode

In general, you want to spend as little of your time in vim’s insert mode as possible, because in that mode it acts like a dumb editor. This is why most vim novices spend so much time in insert mode — it makes vim easy to use. But vim’s real power lies in command mode! You’ll find that the better you know vim, the less time you will spend in insert mode.

Use h, j, k, and l

The first step to efficient editing in vim is to wean yourself from the arrow keys. One of the the advantages of vim’s modal design is that you do not need to constantly move your hands back and forth between the arrow keys and the letter keys; when you are in command mode, the letters h, j, k and l correspond to the directions left, down, up, and right, respectively. It takes some practice to get used to, but you will notice the speed difference once you’re used to it.

When you are editing e-mail or other paragraph-formatted text, you might notice that the direction keys skip more lines than you expect. This is because your paragraphs appear as one long line to vim. Type g before h, j, k or l to move by screen lines instead of virtual lines.

Use motions to move the cursor in the current line

Most editors have only simple commands for moving the cursor (left, up, right, down, to beginning/end of line, etc). vim has very advanced commands for moving the cursor; these commands are referred to as motions. When the cursor moves from one point in the text to another, the text between the points (and including the points themselves) is considered to be “moved over.” (This will be important later.)

Here are a few of the more useful motions:

fx Move the cursor forward to the next occurance of the character x on the current line (obviously, x can be any character you like). This is an extremely useful command. You can type ; to repeat the last f command you gave.
tx Same as above, but moves the cursor to right before the character, not all the way to it. (It’s very useful, really.)
Fx Move the cursor backward to the next occurance of the character x on the current line.
w Move the cursor forward by a word.
b Move the cursor backward by a word.
0 Move the cursor to the beginning of the current line.
^ Move the cursor to the first character on the current line.
$ Move the cursor to the end of the line
) Move the cursor forward to the next sentence. (Useful when editing e-mail or text documents.)
( Move the cursor backward by a sentence.

Move efficiently through the file

vim has many commands that can send you to where you want to go in your file — there’s rarely a need to scroll manually through it. The below keystrokes are not technically motions, since they move around in the file instead of in a particular line.

<C-F> Move the cursor forward by a screenful of text
<C-B> Move the cursor backward by a screenful of text
G Move the cursor to the end of the file
numG Move the cursor line num. (For instance, 10G moves to line 10.)
gg Move the cursor to the beginning of the file
H Move the cursor to the top of the screen.
M Move the cursor to the middle of the screen.
L Move the cursor to the bottom of the screen.
* Read the string under the cursor and go to the next place it appears. (For instance, if your cursor was somewhere on the word “bob,” the cursor would move to the next occurance of “bob” in your file.)
# Same as above, except it moves the cursor to the previous occurance.
/text Starting from the cursor, find the next occurance of the string text and go to it. You will need to press Enter to execute the search. To re-execute your last search, type n (for next occurance).
?text Same as /, but searches in the opposite direction.
ma Make a bookmark named a at the current cursor position. A bookmark can be named any lowercase letter. You can’t see the bookmark, but it’s there!
`a Go to bookmark a. Important: that’s a backtick, not a single quote. The backtick is located to the left of the 1 on most keyboards.
`. Go to the line that you last edited. This is very useful! If you need to scroll through the file to look something up, you can go back to where you were without bookmarking it by using the `. command.

Typing efficiently

Use keyword completion

vim has a very nice keyword completion system. This means that you can type part of a long word, press a key, and have vim finish the word for you. For instance, if you have a variable called iAmALongAndAwkwardVarName somewhere in your code, you probably don’t want to type the whole thing in every time you use it.

To use keyword completion, just type the first few letters of the string (e.g. iAmAL) and press <C-N> (that means hold down Ctrl and type N) or <C-P>. If vim doesn’t give you the word you want at first, keep trying — vim will cycle through all completions it can find.

Enter insert mode intelligently

Most users new to vim get into insert mode by typing i. This works, but it’s often pretty inefficient, since vi has a host of commands that leave the editor in insert mode. Here are some of the more popular ones:

i Insert text to the left of the current character.
I Insert text at the beginning of the current line.
a Insert text to the right of the current character.
A Insert text at the end of the current line.
o Create a new line under the current one and insert text there.
O Create a new line above the current one and insert text there.
c{motion} Delete (change) the text moved over by {motion} and insert text to replace it. For instance, c$ would delete the text from the cursor to the end of the line and enter insert mode. ct! would delete the text from the cursor up to (but not including) the next exclamation mark and enter insert mode. The deleted text is copied to the clipboard and can be pasted.
d{motion} Delete the text moved over by {motion} — same as c{motion}, but doesn’t enter insert mode.

Moving blocks of text efficiently

Use visual selections and the appropriate selection mode

Unlike the original vi, vim allows you to highlight text and perform operations on it. There are three main visual selection modes (that is, text highlighting modes). These modes are as follows:

v Characterwise selection mode. This is the selection mode that most people are used to, so practice with it before trying the others.
V Linewise selection mode. Whole lines are always selected. This is better than characterwise mode when you want to copy or move a group of lines.
<C-V> Blockwise selection mode. Extremely powerful and available in very few other editors. You can select a rectangular block and any text inside that block will be highlighted.

All the usual cusor movement keys apply — so, for instance, vwww would go into visual selection mode and highlight the next three words. Vjj would go into linewise visual selection mode and highlight the current line and the two lines below it.

Cutting and copying from visual selections

Once you have a highlighted selection, you probably want to do something with it. Some of the more useful commands you can give when an area of text is highlighted:

d Cut (delete) the highlighted text and put it into the clipboard.
y Copy (or yank, which is vim-ese for “copy”) the highlighted text into the clipboard.
c Cut the highlighted text into the clipboard. This is just like d, except it leaves the editor in insert mode.

Cutting and copying from non-visual selections

If you know exactly what you want to copy or cut, you can do it without entering visual mode. This saves time.

d{motion} Cut the text moved over by {motion} to the clipboard. For instance, dw would cut a word and dfS would cut from the cursor up to and including the next capital S on the current line of text.
y{motion} Copy the text moved over by {motion}.
c{motion} Cut the text moved over by {motion} and leave the editor in insert mode.
dd Cut the current line.
yy Copy the current line.
cc Cut the current line and leave the editor in insert mode.
D Cut from the cursor to the end of the current line.
Y Yank the whole line, just like yy. (Yes, it’s inconsistent! You can use y$ to do what you would expect Y to do.)
C Cut from the cursor to the end of the current line and leave the editor in insert mode.
x Cut the current character. (This is sort of like a command-mode backspace.)
s Cut the current character and leave the editor in insert mode.


Pasting is easy. Put the cursor where you want the pasted text and type p.

Using multiple clipboards

Most editors have a single clipboard. vim has many more; clipboards in vim are called registers. You can list all the currently defined registers and their contents by typing :reg. Typically, you’ll be using the lowercase letter registers; the others are used for various internal vim purposes and are only occasionally helpful.

To use a specific register for a copy or paste operation, simply type "a before the command for the operation, where a is the register you want to use.

For example, to copy the current line into register k, you could type "kyy. (You could also type V"ky. Why would that work?). That line would stay in register k until you specifically copied something else into register k. You would then use "kp to paste the text from register k.

Avoiding repetition

The amazing . command

In vi, typing . (a period) will repeat the last command you gave. For instance, if your last command was dw (delete word), vi will delete another word.

Using counts

Counts are one of the most powerful and time-saving features of vim. Any command can be preceded by a number. The number will tell vim how many times to execute the command. Here are a few examples:

3j will move the cursor down three lines.

10dd will delete ten lines.

y3f" will yank (copy) text from the cursor to the third quotation mark after the cursor on the current line. Counts are useful to extend the range of a motion in this manner.

Recording macros

Occasionally, you’ll find yourself doing the same thing over and over to blocks of text in your document. vim will let you record an ad-hoc macro to perform the operation.

qregister Start macro recording into the named register. For instance, qa starts recording and puts the macro into register a.
q End recording.
@register Replay the macro stored in the named register. For instance, @a replays the macro in register a.

Keep in mind that macros just record your keystrokes and play them back; they are not magic. Recording macros is almost an art form because there are so many commands that accomplish a given task in vim, and you must carefully select the commands you use while your macro is recording so that they will work in all the places you plan to execute the macro.

Writing code in vim

vim is an excellent editor for source code because it has many features that are specifically designed to help programmers. Here are a few of the more handy ones:

]p Just like p, but it automatically adjusts the indent level of the pasted code to match that of the code you paste into. Try it!
% Putting the cursor on a brace, bracket, or parenthesis and pressing % will send the cursor to the matching brace, bracket, or parenthese. Great for fixing parse problems related to heavily nested blocks of code or logic.
>> Indent the highlighted code. (See the earlier section about efficient text selection. If no text is selected, the current line is indented.)
<< Like >>, but un-indents.
gd Go to the definition (or declaration) of the function or variable under the cursor.
K Go to the man page for the word currently under the cursor. (For instance, if your cursor is currently over the word sleep, you will see the man page for sleep displayed.)

Sail Away Selah’s Way

Sorry for my absence. I’m sailing to Australia at the moment.

Scrobble to with Pianobar

Pianobar is an excellent command-line Pandora client. It has support for calling external commands when songs play, pause, skip, et cetera via the event_command directive. With this, we can add a little scrobbling script and get our listening tracks sent to

First, cd to ~/.config/pianobar (or mkdir if it doesn’t exist) and paste in as, and chmod +x it. You’ll have to open and add your API key (get one from and username/password.

Next, go to pylast and get and put it in the same directoy as

Finally, open or create ~/.config/pianobar/config and put in these lines:

user =
password = somethinglong
event_command = /Users/username/.config/pianobar/

Bonus: If you’re a Pandora One subscriber, you can get the high-quality stream by adding the audio_format = mp3-hifi directive.

(Note: Due to shortcomings in pianobar, your password can’t have funny characters like & in it.)

When you start pianobar, it should start showing up on after the first song finishes. If there are problems then pianobar should print them to stderr.

Edit 2011-06-02: The gist has been updated to support Now Playing. Update your with the new version and pianobar should automatically start updating with Now Playing.

Puppet bootstrap

When starting a brand-new puppet environment, this is the procedure I usually follow to get a client/server bootstrap. First get puppet installed on two boxes, one will be a master and one (or both) will be clients.

(Note: You need some kind of name-resolution infrastructure, either via DNS or /etc/hosts. Puppet clients by default try to contact your master via the short name puppet then the long name

To start, do the following (I’m assuming Puppet 2.6 commands. For 0.25 use puppetd for puppet agent and puppetmasterd for puppet master):

  • On the master: touch /etc/puppet/manifests/site.pp and puppet master --no-daemonize --verbose to watch for connections.
  • On the client: puppet agent --test You should see it generate and submit a cert to the master to be signed.
  • On the master: puppet cert --list to get the list, then puppet cert --sign <certname> to sign it.
  • On the client: puppet agent --test and it should get it’s signed cert, then a catalog from the master if you have a node definition for it.

If that messes up and you want to start over:

  • On the master: puppet cert --clean <certname> to clean the client’s cert.
  • On the agent: puppet agent --configprint ssldir then rm -rf that dir to remove its cert.

Once server<->client is working fine fine, ^C the puppet master and run:

  • On the master: service puppetmaster start
  • On the agent: service puppet start

Now you should be able to put modules in /etc/puppet/modules and node definitions in /etc/puppet/manifests/site.pp and the master will stat these every 2 seconds for changes. You can wait for the client to automatically contact the master for its catalog, or run puppet agent --test to see immediate results.

GTD Gmail Filter Tricks

One of my favorite blog posts of all time on Gmail GTD techniques is this one by Neil Dixon (Edit: Link is dead. Use this blog post instead). Using this technique means that if I want to manage a todo list in combination with my email GTD workflow, I have to check two places to see my task list (even if I’m using the task list built-in to Gmail). How can I combine my email GTD with my todo GTD into one place?

A quick tip about a gmail feature that is not well known. You can append +anything to the end of your username for any Gmail or Google Apps domain and it will still be sent to you. For example, or can be turned into or

To integrate my Gmail GTD workflow with an email-driven todo list, I should theoretically be able to have any single email sent “from me to me and only me” be archived and labeled @action. Gmail filters don’t currently have a way to filter based on “to me and only me,” so I improvised.

Create an email filter for “” (or whatever) to auto-archive and apply your @action label (and optionally a “Todo” label via a second filter), and send an email to this address with your one-line todo sentence in the email subject line. Thereafter, when composing an email you should be able to type “todo” into the To: line and hit <tab> and have it complete for your Todo list.

Protip: If you have keyboard shortcuts enabled and are viewing your @action emails, you can use ‘y’ to remove the @action label without removing the Todo label, essentially marking it “complete.”

Fullscreen Cocoa Emacs on OS X 10.6

The age old battle of editors rages on, but I have a feeling that it’s largely due to misunderstandings of purpose and ignorance of the opposing party. I have been a vi/vim user for 10+ years and found myself strongly advocating vim’s qualities of being lighter and more prevalence than emacs without having so much of an idea of how to quit emacs other than with pkill. Then I found myself with 3 months of free evenings and a slow-to-nonexistent internet connection and decided to fill some of that time and try emacs.

Long story short, I really like emacs for some tasks, and I really like vim for others. I love hacking on my .emacsrc but had been mostly constrained to running it in the terminal over ssh connections until I got a Mac. Now Macs are funny in that they are very GUI-gooey, but they manage to pull it off in a way that doesn’t make me, a hardcore terminal-only tiling-window-manager user, want to take a hammer to it.

Subsequently, I have been using the gui version of Emacs more and more, and it’s really something special on OS X. One feature I feel it’s lacking is fullscreen support. When working with vim, I toggle to fullscreen terminals when I really want to focus, so Emacs should allow me to do the same thing.

Shamelessly adapted from citizen428’s guide, guide, here is the step-by-step way to get fullscreen support written by typester on OS X vanilla emacs without having to rely on someone else’s compiled

# Get the official Emacs repo:
git clone git://

# Add typester as a remote and fetch his branches
git remote add typester git://
git fetch typester

# From a3585f6c2a to typester/feature/fullscreen are the 14
# commits we want to apply to our tree. Create a patch and apply to
# HEAD for the most cutting-edge Emacs
git checkout origin/master
git diff a3585f6c2a typester/feature/fullscreen | patch -p1

# Create a commit so you can just maintain your git tree for later
git commit -a -m "Typester's fullscreen patch" && git clean -d -f

# Alternatively, you could cherry pick all of typester's commits
# and not have to commit anything.
git cherry-pick 329d3eeab6e1997eed97e1fe40e5489a4c68edae \
                592a51bc2938e7c40a3c60b5a5301ea1510a0e74 \
                b9ff1ae683cf453da0415f5c83e329bbff35b9fa \
                000208df01f5c3af7d6f584304bc1a898701f87b \
                d8cf14d0243e5046f77051d05af31af77125d583 \
                df1e25370854fc04e4c217c214e9c02084c1ddae \
                41a9a1d946061c87ab6a6faf484a5b2e5812229f \
                c586de33d8e9ed6c94887526e9891de1788ba3f8 \
                663de471ea537083a86832c46984eab08407be90 \
                d1e183b01bd9f4608203aec5ae06518c37932a98 \
                b549c85507a1f52230c16ace217fe4e66308fc97 \
                5f2a5f555f475d570e6db2ffea4b259f0a90c302 \
                94512b4fd984146a62cde7d9c925058d0ebd0be3 \

# Done with typester's remote
git remote rm typester

# build emacs
./configure --with-ns
make && make install
mv nextstep/ /Applications/

# Later, to update emacs and keep fullscreen just do:
git pull origin master

I then bound “M-RET” to ns-toggle-fullscreen in my .emacsrc.

Suppress iTunes invocation on Pause/Play key

On my Macbook I have the F1-12 keys accessible under the Fn key, and instead by default they have various actions such as changing the volume or brightness, activating exposé, or other various functions. F7-F9 are shortcut keys for Next Track, Play/Pause, and Previous Track for music or other media.

When I got my mac I knew I didn’t want iTunes for various reasons, not the least of which is that it can’t play flac. I’m still looking for a good alternative for heavy usage of listening to my own music, but between‘s app and Pandora‘s awesome console program called pianobar (you can `port install pianobar`, I’ve only been listening to my own music about 50% of the time.

The other 50% of the time I’ve been using VLC, and it works out of the box with the keyboard media shortcuts! Except for one hitch… any time I press the Play/Pause button OS X tries to start iTunes, and there are more than one angry threads about Apple turning a deaf ear to making this configurable/disabled short of deleting iTunes. But there is a hack if you want both iTunes and VLC to co-exist.

Essentially I rename the iTunes binary and replace the original with a small script that first checks if VLC is running, and if not then launches iTunes. Just paste this into Terminal:

cd /Applications/
sudo mv iTunes iTunesX
sudo cat > iTunes << EOF
#!/usr/bin/env python
import sys, os, subprocess
launch = True
blocker = ""
apps = ["Spotify", "Songbird", "VLC"] # Add more blocking programs here
ps = subprocess.Popen("/bin/ps -x", shell=True, stdout=subprocess.PIPE)
for line in"\n"):
    for app in apps:
        if app in line:
            launch = False
            blocker = app
if launch :
    os.spawnvp(os.P_WAIT, '/Applications/', sys.argv)
else :
    print "Not launching iTunes while %s is running." % blocker
sudo chown root:admin iTunes #To match the perms of the other iTunes files
sudo chmod 775 iTunes

If you want to "reset" your iTunes, you can paste this:

cd /Applications/
sudo mv iTunesX iTunes

Dynamic Git-branch Puppet Environments

When working with puppet, eventually you’ll arrive at any one of several conclusions; you have a mass of code and want to refactor it, you have a cool idea that you want to prototype, or you want to add some code but don’t want to push it to production until it’s done. Puppet’s answer to this is environments. The documentation says that you can have arbitrary environments and redefine your configuration parameters in puppet.conf based on which environment you wish your clients to use. The general idea is to have several environments like “Production,” “Development,” and “Testing.”

Now, for me this falls down when I want to break away from more standard development cycles that no longer fit in the dev->test->prod workflow. Or if I have groups of developers working on disjointed projects but don’t want to support completely separate puppetmasters with their own clients. Or when I want to follow git’s mentality of “branching is cheap; branch often.”

I’d like to be able to say `git checkout -b newfeature`, add some resources and templates, `git push`, and instantly use it on my clients. No reconfiguration of the puppetmaster needed. When I finish, `git checkout production ; git merge newfeature ; git branch -d newfeature` will put all aright again.

In puppet.conf I can use $environment to reference the current environment for setting modulepath and manifest, but I don’t actually need the [newfeature] environment declaration section in the puppet.conf. This allows me to create a config thus:

  environment = production
  manifest    = $confdir/environments/$environment/manifests/site.pp
  modulepath  = $confdir/environments/$environment/modules

  environment = production

This says that the master will base the manifest and module path on the environment, which is passed by the agent to the master. If I changed the [agent] section on a puppet agent to have environment = newfeature instead, then the catalog compiled by the master would be from /etc/puppet/environments/newfeature. If an agent does not pass the environment variable then the default production will be used. (Big Note is that your files and templates really only work cross-environment if you’re using modules. You can either clone all of your modules per environment, or you can have some modules be per environment and some be global. Check out Volcane’s blogpost on using your modules in this manner.)

Now that I have puppet set up to take arbitrary environments, lets make this useful. To implement it so that git branches are recognized as separate environments, I have to keep your puppet manifests and modules in git. (You’re already doing that, right?) Starting without any environments, I’ll rearrage my puppet configs to look like this:

     | puppet.conf
     | fileserver.conf
     - environments/
       - production/
         + manifests/
         + modules/

In the production directory I `git init ; git add .` (then considering that I have a gitosis instance set up and an ssh key ready to be used by the puppet user,) `git remote add origin ; git push origin master`. Now for the git hook I go open puppet-environments.git/hooks/post-receive and put this bourne code in, configured with my git repo url and the key by which my puppet master can pull from gitosis:

read oldrev newrev refname

BRANCH=`echo $refname | sed -n 's/^refs\/heads\///p'`
SSH_ARGS="-i /var/lib/puppet/.ssh/id_rsa"

if [ "$newrev" -eq 0 ] 2> /dev/null ; then
  # branch is being deleted
  echo "Deleting remote branch $BRANCH_DIR/$BRANCH"
  ssh $SSH_ARGS $SSH_DEST /bin/sh <<-EOF
    cd $BRANCH_DIR && rm -rf $BRANCH
  # branch is being updated
  echo "Updating remote branch $BRANCH_DIR/$BRANCH"
  ssh  $SSH_ARGS $SSH_DEST /bin/sh <<-EOF
    { cd $BRANCH_DIR/$BRANCH && git pull origin $BRANCH ; } \
    || { mkdir -p $BRANCH_DIR && cd $BRANCH_DIR \
         && git clone $REPO $BRANCH && cd $BRANCH \
         && git checkout -b $BRANCH origin/$BRANCH ; }

Now after I’ve cloned puppet-environment.git to my laptop for hacking, I can `git checkout -b newfeature` and end up with an exact representation of my production code, but free to be hacked on until it’s ready to be pushed and tested. Pushing will instantiate it in a new environment. After I’ve pushed, I can run `puppet agent --test --environment newfeature` on a puppet client and get a one-time run inside your newfeature environment, and it will revert back to production after that run.

Finally, what if I want my hosts to remember which --environment I last ran on it? Or how about being able to configure the nodes’ environment right from puppet? For that I can make a small module with a parameterised class called “environment.” Here’s my init.pp:

class environment($env = $environment) {
  file { '/etc/puppet/puppet.conf':
    owner   => 'puppet',
    group   => 'puppet',
    mode    => '644',
    content => template("environment/puppet.conf.erb"),

As the final step, I copy my clients’ puppet.conf to modules/environment/files/puppet.conf.erb and replace the environment = production section in [agent] with environment = <%= env %>.

Now I can add include environment in my base node or any node that you want to “remember” it’s environment given via --environment. If I’d like to configure a node to always use a specific environment then I can instead use class { environment: env => "production" } and it will always revert to the production environment.

Thanks to Marut and the guys at PSU for the inception!

Joining the Tautology Club

Hunter Haugen is a sys-admin/hacker of 10 years turned DevOps in an attempt to understand, play, and ultimately win The Game. Sorry, I mean The Game. My current battleground is a little open source company called Puppet Labs in the home of open source, Portland Oregon. My theatre is the world (it’s in my job description) — as of last Monday I have signed on as a consultant with Puppet’s Professional Services team to offer training seminars around the world for those who want to understand and use the configuration management software called ‘Puppet.’ I also have aspirations of being a successful entrepreneur one day, but so far haven’t gotten the hang of the ‘successful’ part.

As with all good Portlanders, I relish the experience of engaging in the most efficient form of mechanized transportation and often go on long (60-600 mile) bicycle rides with my family of 6 (3 are siblings). I also enjoy the motorized version, which is also my primary form of transportation at the moment, rain or shine. Quick trivia: I was run over by a tractor while driving a motorcycle in India. My other interests are keyboards, ultralight backpacking, languages, and a collection of things that a large percentage of the population also find captivating.

Coming up next: technical and non-technical things about OS X, productivity, coding, and things relating to but not entirely 42.